Add 35-Day Filter for PR Stats on Miner Details Page

[bittoby]

Description

The miner details page shows PR statistics (merged, open, closed) based on full history instead of the 35-day scoring window. This makes the numbers inconsistent with what the scoring system actually evaluates.

Steps to Reproduce

1. Go to the miner details page for any miner
2. Look at the PR counts in the Overview tab (credibility tile, donut chart)
3. Switch to the Pull Requests tab and check the status counts
4. Compare these numbers with what the 35-day scoring window should reflect

Expected Behavior

Users should be able to view PR stats scoped to the 35-day lookback window that the scoring system uses. There should be a filter option to switch between full history and the 35-day window so the numbers match the actual evaluation period.

Actual Behavior

All PR counts (merged, open, closed) show full history with no way to filter by the 35-day scoring window. The credibility chart, score card, and PR table all display all-time totals.

Environment

• Browser: All
• OS: All

Additional Context

The individual PR records with timestamps (prCreatedAt) are already available client-side, so the filtering can be handled on the frontend.

The default view can remain as full history, with an option to toggle to the 35-day window.

[plind-dm]

I think it's sth should be done on the backend side 🤔

[bittoby]

I mean the UI should have a 35-day window filter option, not just full history.

[plind-dm]

Yeah, there're some non-sense glitches on new version

[statxc]

@bittoby If you still don't start this issue, can I work on this? tbh I just found same problem with you 😉

[bittoby]

Hi, @statxc Thanks for interesting this issue.
But to solve this issue, I think it will need a small change in the gittensor repo.
With the current UI, we can filter All, Open, and Merged PRs, but not Closed PRs, because the backend doesn’t return a closed_at field. At least that’s how it looks to me.
Can you double-check on your end? I would appreciate your thoughts.