Add per-repo failure tracking to refresh scheduler

[garvitkaushik-123]

Summary

• Add RepoFailureTracker class that counts consecutive failures per repo across refresh cycles
• Log warning when a repo fails 3+ consecutive times, critical at 5+
• Log recovery when a previously failing repo succeeds
• After each cycle, summarize all repos with repeated failures in the status log
• Backwards compatible: failure tracker is optional (defaults to None if not passed)

This addresses the "repeating fails should trigger some kind of warning" requirement from #15.

Partial fix for #15

Test plan

• Verify scheduler starts normally with no behavioral change when all repos succeed
• Simulate a repo failure (e.g., invalid repo name) and verify warning logs appear after 3 consecutive failures
• Verify the failing repo summary is logged at the end of each cycle
• Verify recovery is logged when a previously failing repo succeeds

🤖 Generated with Claude Code

[partvishegy]

Hey @garvitkaushik-123! I appreciate the effort, but as you can see this issue is heavily underspecified at the moment. Althought the repo is open, I did not expect contirbutors yet. I do have some architectural rethinking to do, and the issues are more like half-thoughts on a todo list, then specs. #15 is a good example of this: the scheduling part is essential for this project, but in its current form patching it does not create much value. I do have a plan to redo it, but it is not recorded here. As you must know: if the business requirements are not defined clearly, the outcome can hardly be correct.

If you are serious about contributing to this project, let me know, and I am happy to put more effort into detailed specs on these issues, and the other unrecorded ones.

I can see that you added a nice Test plan! Have you done any testing yourself? Did you try running the project? How did it go?