explorer: sync #facetNote on URL deep-link + mode transitions (#234 step 1)

[rdhyee]

Summary

Roadmap step 1 of #234 (filter-semantics architectural direction). Pure bug fix, no behavior change beyond making the cluster-mode honesty note (#facetNote) honest in two cases where it had drifted out of sync.

• URL deep-link: arriving at /explorer.html?material=… restored checkbox state via setCheckedValues, which doesn't fire the change event, so #facetNote stayed hidden even with active filters. A user opening a shared URL saw dimmed cluster dots with no explanation.
• Mode transitions: the note was only updated inside handleFacetFilterChange. A user in cluster mode with the note showing kept seeing it after zooming to point mode (and vice versa).

Fix: extract syncFacetNote() as the single source of truth for the visibility invariant

visible ⇔ (any facet active) ∧ (mode === 'cluster')

and call it from the four state-mutation sites:

• facetFilters cell (URL deep-link restore)
• handleFacetFilterChange (user toggles a facet checkbox)
• enterPointMode / exitPointMode (mode transitions)

Test plan

• New spec tests/playwright/facetnote-url-load.spec.js:
  • cluster mode + ?material=<uri> → #facetNote visible
  • cluster mode + no facet params → #facetNote hidden (negative control)
• Fault-injection: confirmed the new spec FAILS against pre-fix code (expected: 'block', received: 'none') and PASSES against the fix.
• Full tests/playwright/url-roundtrip.spec.js (5 tests, includes point-mode deep-link and cross-context round-trip) passes — no regression.
• Manual: open /explorer.html?material=<any URI> and confirm note appears under the cluster legend.

Notes for review

This PR is intentionally scoped to step 1 of the #234 roadmap (the 1–2 hour bug fix). It does not touch:

• applyQueryToFacetFilters itself (the URL-param read path is unchanged)
• The refreshFacetCounts cross-filter live-query path (B1 work, step 3)
• The cluster-mode H3-summary-vs-filter honesty story (C3 work, step 8)

The roadmap synthesis with sign-off and revised sequencing is in issue #234.

🤖 Generated with Claude Code

[rdhyee]

Code review result: no code-level issues found in this PR.

I checked the PR patch around the centralized syncFacetNote() logic and the new Playwright regression spec. The implementation matches the intended invariant: facet note visible only when a facet filter is active and Explorer is in cluster mode.

Validation run locally:

• quarto render explorer.qmd passed
• git diff --check upstream/main...HEAD passed
• CI=1 ./node_modules/.bin/playwright test tests/playwright/facetnote-url-load.spec.js tests/playwright/url-roundtrip.spec.js --project=chromium --reporter=line passed: 7 passed in 1.3m

Merge blocker: GitHub currently reports mergeable=CONFLICTING / mergeStateStatus=DIRTY, and git merge-tree HEAD upstream/main confirms a content conflict in explorer.qmd. This needs a rebase or conflict resolution before merge.

[rdhyee]

Rebased onto upstream/main (was branched off a stale fork-main). Single conflict was in exitPointMode: upstream replaced the cachedBounds = null; … cache-invalidation block with a ++requestId stale-guard (#221 round 2). Resolution kept upstream's version and inserted syncFacetNote() right after the pushState call, before the requestId bump.

Re-ran both suites against the rebased commit (230e763):

✓ tests/playwright/facetnote-url-load.spec.js (2 tests)
✓ tests/playwright/url-roundtrip.spec.js (5 tests)
7 passed (1.3m)

mergeable=MERGEABLE now per gh pr view. Thanks for the catch, codex.