GH-5268: fix(briefs): receipts digest — record SentAt = query End to close the sub-second (End, SentAt) skip window

[alekspetrov]

Summary

Automated PR created by Pilot for task GH-5268.

Closes #5268

Changes

GitHub Issue GH-5268: fix(briefs): receipts digest — record SentAt = query End to close the sub-second (End, SentAt) skip window

Problem

Follow-up from PR#5258 re-review (note-level, non-blocking): in ReceiptsScheduler.runDigest (internal/briefs/receipts.go), the executions query uses End: now captured at the top of the function, but after delivery the brief_history record is stamped with a fresh time.Now():

record := &memory.BriefRecord{
    SentAt:    time.Now(),   // later than the query's End by delivery latency
    ...
}

The next digest windows [lastRecord.SentAt, now). A run whose completed_at lands in the gap (End, SentAt) — i.e. completes during the ~100–500ms of Telegram delivery — is after this digest's window and before the next one's start, so it is never receipted. Breaks the exactly-once guarantee established by GH-5261, just at sub-second scale.

Fix

One line: stamp the record with the same now used as the query End (SentAt: now), so consecutive windows tile exactly: [prev.End, now).

Acceptance

1. runDigest records SentAt equal to the query's End bound (same time.Time value).
2. Test: an execution with completed_at exactly between the first digest's query End and its (formerly later) send-completion time appears in the second digest — extend TestReceiptsSchedulerRunNow_InFlightRunAppearsInNextDigest or add a sibling.
3. make lint && make test green; no other behavior change.

Refs

• PR#5258 re-review note: GH-5257: feat(briefs): daily receipts digest — per-run cost receipt lines at configurable time (default 18:00) #5258 (comment)
• fix(briefs): PR#5258 revision — receipts digest window drops in-flight/post-18:00 runs permanently #5261 (the window redesign this completes)

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[alekspetrov]

Post-merge review — APPROVE

Exactly the requested fix, nothing more: SentAt stamped with the same now used as the executions query's End bound, so consecutive digest windows tile as [prev.End, now) with no sub-second gap.

Test quality is the highlight: TestReceiptsSchedulerRunNow_SentAtMatchesQueryEnd reproduces the actual race — the mock sender's onSend hook inserts a completing execution synchronously mid-SendBriefMessage, i.e. strictly after the query ran and strictly before SentAt is stamped. That's the exact (End, SentAt) dead window from GH-5268, and the test proves the row surfaces in digest 2. Boundary semantics verified: query is completed_at >= start AND < End, so an exactly-at-End row lands in the next digest only — no gap, no duplicate.

All 3 acceptance criteria met; 8/8 checks green; the only non-receipts change is the additive optional onSend hook on the shared mock. No follow-ups.