the update job still fetches repositories from github one at a time. after DB writes are batched, github/network latency will likely become the next major bottleneck.
- add a small worker pool for github repository reads, likely 5-10 workers
- preserve existing 404 delete handling and error counts
- handle primary and secondary github rate limits with backoff
- keep database writes batched after results are collected
- verify update duration and github error behavior in ecs logs
the update job still fetches repositories from github one at a time. after DB writes are batched, github/network latency will likely become the next major bottleneck.