Org assignment history, mobile support, and a correctness pass

Data model
- employee_assignments records org placement over time (valid_from/valid_to),
  written by a trigger on `employees` rather than inside each RPC: ~70
  `update employees` statements spread over fifteen migrations mean per-call
  bookkeeping would miss paths today and again with every future RPC. A
  partial unique index enforces the one-open-interval invariant the trigger
  relies on when closing the current row.
- The Organigramm gains a Stichtag (default today). Membership comes from
  entry/exit/karenz, past placement from the new history, future placement
  projected from pending_org_changes. Placements predating the migration are
  backfilled with today's values and flagged as such in the UI, since
  employee_history only ever stored free text and cannot be reconstructed.

Correctness
- Reports and exports silently truncated at PostgREST's 1000-row cap
  (db.max_rows); employee_history is already past it at ~800 staff. Every
  whole-table read now pages explicitly.
- XLSX date cells were a day early: ExcelJS converts a Date to an Excel
  serial straight off getTime(), so a Date built at local midnight lands on
  the previous day's serial in any positive-offset zone.
- Date handling is pinned to Europe/Vienna throughout, and date-only strings
  are formatted without a Date round-trip. The dashboard's YTD window was
  built by round-tripping a local Date through toISOString(), which shifted
  it a day early and dropped 31 December entirely.
- Export routes parsed measure/group/split/eventType with unchecked `as`
  casts, so an unknown value reached column headers as `undefined` and the
  Content-Disposition filename. Parsed against the label maps now, with the
  filename slugged as a backstop.
- toXlsx keyed columns by header text, silently dropping the second of any
  two columns sharing a name — split columns take their header from data.
- The org chart tree walks had no cycle guard; nothing in the schema forbids
  a manager_id cycle, and one would hang the tab rather than misreport.
- The login page reflected ?error= verbatim, letting anyone put arbitrary
  text on the real sign-in screen; messages are looked up by code now.
- React Flow needs elementsSelectable on, or it sets pointer-events:none on
  the whole node and the expand control stops responding.

UI
- Mobile: the shell was unusable below lg — a fixed 236px margin pushed
  content off-screen with no mobile navigation at all. The sidebar is now a
  drawer, dvh replaces vh, safe-area insets are honoured, inputs are 16px so
  iOS stops zooming on focus, and form grids stack.
- Org chart nodes redesigned: per-kind accent stripes and icons, vacant
  roles called out, expand control moved to the bottom edge carrying the
  child count.
- Pagination is windowed; it previously rendered one link per page (54 for
  the employee list, unbounded for the audit log).
- Positions page reduced to open positions with a single "Besetzen" action.
- The employee Organisation tab links into the org chart focused on that
  person, reusing the chart's existing search-match highlighting.

Also included, uncommitted until now
- Dependants, HR notes, academic titles, split address fields, position
  validity and role/employment fields, with their migrations and UI.
- Docker/compose deployment setup, data-model and security-review docs.

supabase/migrations/20260724120000_employee_assignment_history.sql

@@ -0,0 +1,120 @@
+-- Org-assignment history, so the Organigramm can be shown as of any date.
+--
+-- Until now `employees` carried only the *current* placement (manager_id,
+-- team_id, division_id, job_title, is_lead, org_level) and every mutation
+-- overwrote it in place. employee_history recorded that something happened,
+-- but only as free text — no old/new values — so a past reporting line was
+-- gone for good. Forward-looking dates already worked (pending_org_changes
+-- holds not-yet-due changes structurally); it was the past that could not be
+-- reconstructed. This migration adds the missing timeline.
+--
+-- Captured by a trigger rather than by editing the mutating RPCs: there are
+-- ~70 `update employees` statements spread over fifteen migrations (several
+-- of which redefine the same function repeatedly), so per-RPC bookkeeping
+-- would miss paths today and again with every future RPC. A trigger on the
+-- table catches all of them, including ones not written yet.
+
+create table employee_assignments (
+  id uuid primary key default gen_random_uuid(),
+  employee_id uuid not null references employees(id) on delete cascade,
+  manager_id uuid references employees(id) on delete set null,
+  team_id uuid references teams(id),
+  division_id uuid not null references divisions(id),
+  job_title text not null,
+  is_lead boolean not null default false,
+  org_level int not null,
+  valid_from date not null,
+  -- Exclusive upper bound; null means "still in force". Note this tracks
+  -- *placement*, not employment: the row of someone who has left stays open,
+  -- because whether they were employed on a given date is derived separately
+  -- from entry/exit/karenz dates. Keeping the two apart is what lets a
+  -- rehire reuse the same open row instead of needing it reopened.
+  valid_to date,
+  created_at timestamptz not null default now(),
+  constraint chk_assignment_range check (valid_to is null or valid_to > valid_from)
+);
+
+create index on employee_assignments (employee_id, valid_from desc);
+create index on employee_assignments (valid_from);
+-- At most one open interval per employee — the invariant the trigger relies
+-- on when it looks up "the current row" to close.
+create unique index employee_assignments_one_open on employee_assignments (employee_id) where valid_to is null;
+
+alter table employee_assignments enable row level security;
+create policy "employee_assignments_hr_read" on employee_assignments for select using (is_hr_user());
+
+-- ── Backfill ───────────────────────────────────────────────────────
+-- One open interval per employee, starting at their entry date, holding
+-- today's placement. Changes made *before* this migration are not
+-- recoverable — employee_history never stored structured old values — so a
+-- Stichtag before today's date shows the placement as it stands now for
+-- anyone whose assignment predates this table. Everything from here on is
+-- exact.
+insert into employee_assignments (employee_id, manager_id, team_id, division_id, job_title, is_lead, org_level, valid_from)
+select id, manager_id, team_id, division_id, job_title, is_lead, org_level, entry_date
+from employees;
+
+-- ── Trigger ────────────────────────────────────────────────────────
+-- The date a change takes effect is normally the day it is written: an RPC
+-- with a future effective date does not touch `employees` at all (it queues
+-- into pending_org_changes, and apply_due_pending_changes writes it on the
+-- due date), and one dated today writes immediately. A *backdated* change is
+-- the exception — it writes immediately although it should take effect
+-- earlier — so callers may set `app.effective_date` for the transaction to
+-- say so explicitly; unset, it falls back to the current date.
+create or replace function fn_track_employee_assignment()
+returns trigger
+language plpgsql
+security definer
+set search_path = public
+as $$
+declare
+  v_date date := coalesce(nullif(current_setting('app.effective_date', true), '')::date, current_date);
+  v_open_from date;
+begin
+  if tg_op = 'INSERT' then
+    insert into employee_assignments (employee_id, manager_id, team_id, division_id, job_title, is_lead, org_level, valid_from)
+      values (new.id, new.manager_id, new.team_id, new.division_id, new.job_title, new.is_lead, new.org_level,
+              coalesce(new.entry_date, v_date));
+    return new;
+  end if;
+
+  if new.manager_id is not distinct from old.manager_id
+     and new.team_id is not distinct from old.team_id
+     and new.division_id is not distinct from old.division_id
+     and new.job_title is not distinct from old.job_title
+     and new.is_lead is not distinct from old.is_lead
+     and new.org_level is not distinct from old.org_level then
+    return new;
+  end if;
+
+  select valid_from into v_open_from
+    from employee_assignments where employee_id = new.id and valid_to is null;
+
+  if v_open_from is null then
+    insert into employee_assignments (employee_id, manager_id, team_id, division_id, job_title, is_lead, org_level, valid_from)
+      values (new.id, new.manager_id, new.team_id, new.division_id, new.job_title, new.is_lead, new.org_level, v_date);
+  elsif v_date <= v_open_from then
+    -- Two changes on the same day (or a backdate landing inside the open
+    -- interval): overwrite in place, so no zero-length or inverted interval
+    -- is ever stored.
+    update employee_assignments
+      set manager_id = new.manager_id, team_id = new.team_id, division_id = new.division_id,
+          job_title = new.job_title, is_lead = new.is_lead, org_level = new.org_level
+      where employee_id = new.id and valid_to is null;
+  else
+    update employee_assignments set valid_to = v_date where employee_id = new.id and valid_to is null;
+    insert into employee_assignments (employee_id, manager_id, team_id, division_id, job_title, is_lead, org_level, valid_from)
+      values (new.id, new.manager_id, new.team_id, new.division_id, new.job_title, new.is_lead, new.org_level, v_date);
+  end if;
+
+  return new;
+end;
+$$;
+
+drop trigger if exists trg_track_employee_assignment on employees;
+create trigger trg_track_employee_assignment
+after insert or update on employees
+for each row execute function fn_track_employee_assignment();
+
+grant all on table employee_assignments to anon, authenticated, service_role;