fix(backfill): Monolithic backfill stores PR-inclusive count as openIssuesCount

[glorysr1209-png]

Summary

Closes #176.

Use filtered issue count (exclude PR-shaped /issues rows) for openIssuesCount and backfill return payload.

Changes

• src/github/backfill.ts: openIssueCount = issues.length after PR filter.

Test plan

• npm test -- backfill.test.ts
• Manual: backfill repo with mixed /issues page; openIssues matches true open issues only.

[JSONbored]

@glorysr1209-png the direction makes sense, but this needs tests and a full CI run before I can take it.

A few notes:

• Filtering /issues results before storing openIssuesCount is the right idea. GitHub’s REST issues endpoint includes pull requests, so using the raw fetched count can overstate issue totals.
• This only changes the count assignment; it does not add a regression test proving PR-shaped issue rows are excluded from repo_sync_state, snapshots, and the returned backfill result.
• I also do not see the normal CI validate check on this PR yet, so it is not mergeable even if the patch is small.

Required changes:

• Add a focused regression test where the REST issues page contains both real issues and PR-shaped entries.
• Assert openIssuesCount stores only real issues while openPullRequestsCount still comes from the PR endpoint.
• Get the normal validate workflow green.

Validation expected:

• npm run typecheck
• npm run test:coverage

[JSONbored]

@glorysr1209-png this is still blocked by CI.

A few notes:

• The open issue count bug is in the right backfill area.
• Current validate fails test/unit/backfill.test.ts: expected openIssues: 201, received 1.
• The PR body also needs current validation/safety evidence.

Required changes:

• Fix the failing backfill assertion without weakening the intended behavior.
• Update the PR body to the current template.

Validation expected:

• Focused test/unit/backfill.test.ts
• Full validate pipeline

[JSONbored]

This is stale, closing - please resubmit with all issues addressed.