PerformSync has no timeout, a hung sync blocks all sources

There is no per-sync context deadline. PerformSync receives only the coordinator's base context, which is only cancelled on shutdown. A hung network call (e.g., an API source or git clone that never times out) will block PerformSync indefinitely. Because the coordinator loop is synchronous, this blocks all other sources from syncing for the duration of the hang.

A configurable per-sync timeout (e.g., sync.syncTimeout) should be applied as a context deadline before calling PerformSync, ensuring the deferred status update always fires within a bounded time. The original context from the coordinator handles the cleanup procedure, so this must also keep working after this fix.

Start at the coordinator loop and the PerformSync call, then trace how the base context is used for synchronization and cleanup. Add the configurable sync.syncTimeout deadline while preserving the original coordinator context for cleanup. Done means a hung source cannot block subsequent syncs indefinitely and the deferred status update still fires within the timeout.