feat(scoring): add noise-threshold to cap per-rule findings

[hummbl-dev]

Summary

• Adds filter_noise() to scoring.py — caps findings per rule_id at a configurable threshold
• Adds --noise-threshold flag to all 6 scoring subcommands (analyze, score, score-url, diff, audit-fleet, github-top)
• Reports suppressed findings count when active
• Addresses 10 of 20 open gaps from the github-top audit run (false_positive_candidate class)

Usage

arbiter score /path/to/repo --noise-threshold 50    # cap each rule at 50 findings
arbiter github-top --noise-threshold 30             # cleaner fleet scoring

Test plan

• No filtering when disabled (None)
• Threshold zero disables
• Caps at threshold per rule
• Multiple rules independently capped
• Returns suppressed counts dict
• Rules below threshold unaffected
• Preserves finding order
• Full suite: 257 passed

🤖 Generated with Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ac5d91f8ce

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Reject negative noise thresholds

filter_noise treats any non-zero integer as enabled, so a negative --noise-threshold is accepted and causes every finding to be dropped (counts[...] <= threshold is never true). This produces artificially perfect scores and an incorrect suppressed count (total - threshold exceeds actual findings), which can silently corrupt CI gating if a typo or bad config passes -1; the new CLI flags do not validate this input before calling this function.

Useful? React with 👍 / 👎.