fix(context): score every snapshot entry before applying the candidate cap

[GQAdonis]

Summary

pk context used to cut each scope to ceil(max_candidates / scopes) entries in snapshot order before scoring them. A large KB was therefore mostly invisible to recall:

• with 200 entries and default flags, only 128 were scored, and a match that sorted late was never returned;
• a scope that failed kept its share of the budget, so the other scopes could not use it.

Change (pk-cli/src/main.rs::run_context):

• Every entry of every readable scope is now scored.
• --max-candidates caps the merged, ranked list after de-duplication.
• The ranking is deterministic: score descending, then scope priority, then id.
• The JSON output gains scored_count. candidate_count now means "ranked candidates kept after the cap".

The existing test candidate_budget_is_shared_across_requested_scopes asserted the old count of inspected entries. It now asserts scored_count = 5 and candidate_count = 1.

This is part of the team-aware learning work, plan PR A1 (skill-pack docs/plans/team-aware-learning-memory-implementation.md).

Test plan

• cargo test -p pk-cli --test context_scoring passes. It covers a 200-entry scope with the last-by-id match recalled, a failed scope that does not reserve budget, and byte-identical output across runs with the stated ordering.
• cargo test -p pk-cli --test context passes.
• Negative control: the original code, run on the same 200-entry fixture, returns results=[] with candidate_count=128.

🤖 Generated with Claude Code

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.