fix: brief timing double-send + DB deadlocks; add minute/day scheduling

[Aarav261]

Brief send-time was unreliable and could fire at two times a day.

Root causes:

• The popup auto re-link on every open sent brief_hour (defaulting to 8), silently overwriting a custom hour in the DB.
• schedule_brief used reschedule_job against the persistent jobstore, so a stale trigger left by an overlapping scheduler could survive alongside the new one — firing the brief at both the old and new time.

Fixes:

• Stop the clobber: brief send-time fields are optional in /link-ontrack and preserved from the DB when absent; the auto re-link no longer sends them.
• Make scheduling idempotent (remove-then-add → exactly one job per user) and reconcile orphan brief_* jobs on startup.

Feature — fully customizable send time:

• New brief_minute + brief_dow columns (migrated on both PG and SQLite).
• Settings UI: native HH:MM time picker + Mon-Sun day checkboxes.
• schedule_brief honours hour, minute and day_of_week.

Deadlock fix:

• The contended user row was updated by several separate transactions per request under concurrent extension traffic. Fold the Settings-panel writes into a single update_brief_prefs UPDATE, and add a retry-on-deadlock wrapper (Postgres 40P01/40001, SQLite "database is locked") on the write helpers.

Summary

Related issue

Type of change

• Bug fix
• New feature
• Refactor / cleanup
• Documentation
• Other:

Checklist

• My branch is based on the latest main
• ruff check . passes
• pytest passes (if backend changes)
• npm run lint passes in extension/ (if extension changes)
• I added/updated tests where it makes sense
• I did not commit secrets, .env files, tokens, or database files

Notes for reviewers

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.