This patch prevents one unresponsive shop API request from blocking the production crawler and every later catalog publication.
- Added a browser-side abort deadline to replayed shop API requests, using the crawler's existing
--timeoutvalue. - Kept timeout failures on the existing
network_errorpath so later shops continue and the current retry policy remains authoritative. - Removed stale Chromium singleton symlinks before browser refresh startup and extended the inventory service budget to one hour so atomic publication can complete after scanning.
- Aligned release metadata on
3.7.6.
- Crawler self-test and pytest coverage must pass.
- API tests and Web typecheck must pass before tagging.
- No database migration is required.
- Follow
docs/QUICK_DEPLOY.mdand deploy only thev3.7.6Tag after CI succeeds. - Rebuild the Crawler image and complete one full multi-source publication before restoring production timers.