Add human-reviewed GitHub inspection pipeline

[leadingproblemsolver]

What changed

• added a standard-library GitHub issue discovery and scoring script
• mapped LCE's implemented capabilities to precise issue-search language
• added deterministic relevance, intent, recency, validation, and spam scoring
• added JSON and Markdown review-queue outputs
• added a twice-weekly/manual GitHub Action that uploads the queue as an artifact
• added unit coverage for search construction and qualification behavior
• documented the value-first response gate, inspection-readiness gate, and controlled weekly operating target

Why

The repository needs a repeatable path from its actual mechanisms—source-linked operational memory, deterministic ingestion, context packs, SQLite, CLI, and read-only API—to engineers already experiencing context loss, provenance gaps, stale state, and handoff failures.

The pipeline intentionally stops before public engagement. It produces a human review queue rather than automated comments, protecting maintainer trust and forcing technical relevance before any repository mention.

Impact

This creates an inspectable acquisition loop:

LCE capability → live issue → qualification → human technical help → optional contextual repo reference → repository inspection → feedback/issue/PR/star.

Validation

• branch is 7 commits ahead of main and contains six scoped file changes
• unit tests cover query safety filters, high-signal qualification, and bot/stale penalties
• workflow uses Python 3.12 and only the standard library

Runtime tests were not executed in this environment because GitHub CLI/network-backed checkout tooling was unavailable; the PR is left as draft for CI and human review.