Surface supporting evidence (authors, matched commits, time gaps) for cross-repo co-change rows

[RaghavChamadiya]

Background

Follow-up to #475. The cross-repo co-change page now frames itself honestly as a temporal, same-author hint (#481) and shows frequency and last_date per pair. A remaining ask from #475 is to show why each row exists: the actual commits, authors, and time gaps behind a co-change pair, so a user can judge whether a relationship is meaningful instead of trusting a bare score.

Proposed change

For each co-change pair, surface supporting evidence on demand (for example an expandable row or an inspector panel):

• The authors involved.
• A few example matched commit pairs (the cross-repo commit pair that created the relationship), with short SHAs and dates.
• The time gap between the paired commits.

Why this is not a "good first issue"

The current pipeline does not persist this. detect_cross_repo_co_changes in packages/core/src/repowise/core/workspace/cross_repo.py accumulates strength, frequency, and last_date per file pair, but discards the underlying commit identities and authors. Delivering this needs:

1. Core: capture and persist a bounded sample of evidence per pair (author email, commit SHAs, timestamps) into CrossRepoCoChange / the overlay JSON, with a cap so the artifact does not blow up on large workspaces.
2. Server: expose the evidence through the workspace co-changes endpoint.
3. Types + UI: extend WorkspaceCoChangeEntry and render the evidence.

Pointers

• Core: packages/core/src/repowise/core/workspace/cross_repo.py (detect_cross_repo_co_changes, CrossRepoCoChange, CrossRepoOverlay).
• Type: WorkspaceCoChangeEntry in packages/types/src/workspace.ts.
• UI: packages/ui/src/workspace/co-change-table.tsx.

Acceptance

• A co-change row can reveal its supporting evidence (authors, example matched commits, time gaps).
• The persisted overlay stays bounded in size (evidence is sampled/capped).
• Tests cover the new evidence capture in core.

[rubinashaik2022]

Hi, I’d like to give this a try. Could you assign this issue to me?