fix: prevent stale re-renders of #each items during sequence update (#468)

[dupontbertrand]

Summary

Fixes #468

When the data source of a parent template changes, Tracker can re-run helpers inside #each item views before ObserveSequence has had a chance to diff and remove stale items. This causes items to briefly render with inconsistent data — e.g., an item's msg shows "foo" while the parent data context already says "bar".

Root cause

During Tracker.flush(), autoruns are re-run in invalidation order. When a parent data context changes:

1. Helper autoruns inside existing #each item views are invalidated (via bindDataContext → Blaze.getData() → dataVar.get())
2. The ObserveSequence autorun is also invalidated (it depends on the sequence function)
3. The item helper autoruns can re-run BEFORE ObserveSequence diffs and removes stale items
4. Result: stale item views re-render with old item data but new parent data

Fix

Freeze item views during sequence transitions using 3 hooks:

• onInvalidate (new callback in ObserveSequence.observe): called immediately when the sequence source is invalidated, BEFORE Tracker.flush() re-runs other autoruns. Marks all current item views with _eachItemPendingUpdate = true.
• afterDiff (new callback): clears the flag on surviving item views after the diff is applied.
• doRender guard (in _materializeView): skips re-render if the view or any ancestor has _eachItemPendingUpdate set.

This ensures stale item views never re-render — they are either destroyed by removedAt or updated by changedAt, and the flag is cleared afterward.

Files changed

File	Change
packages/observe-sequence/observe_sequence.js	Add onInvalidate, beforeDiff, afterDiff callback hooks
packages/blaze/builtins.js	Implement hooks in Blaze.Each to freeze/unfreeze item views
packages/blaze/view.js	Guard in doRender to skip re-render when pending update

Test plan

• Tested with 5 scenarios in a dedicated test app:
  1. Different IDs (removedAt + addedAt) — original bug report
  2. Same ID, changed content (changedAt)
  3. Multiple items reduced to one (3 removedAt)
  4. New-style #each item in items
  5. Nested #each (outer + inner)
• All 5 scenarios produce zero stale renders
• Full Tinytest suite: 416/416 passed, 0 failures