Fix excluded-path change detection for pushes

[konard]

Summary

• filter ignored paths before computing the sole job-gating detector output
• use the same detector result for pull requests and direct pushes
• skip lint, test, and package build jobs for excluded-only automatic changes while preserving manual release dispatches
• remove unused detector outputs and constrain Ruff to the compatible 0.8 release line

Root cause

The release workflow treated every push as change-bearing before consulting
detect_code_changes.py. The detector also calculated extension-specific
outputs from the raw file list, so files under excluded directories could
activate jobs on pull requests.

Reproduction

Commit and push a file such as experiments/repro.mjs, dev/log/trace.py, or
docs/case-studies/audit.py. Before this change, one or more change-gated jobs
ran despite the path exclusion.

Verification

• Added an event-matrix regression test for pull requests and pushes across
  experiments/, dev/log/, and docs/case-studies/
• Added workflow policy tests that reject unconditional push gates and unused
  detector outputs
• ruff check .
• ruff format --check .
• mypy src
• python scripts/check_file_size.py
• pytest (61 passed)

Fixes #40

[konard]

Working session summary

Implemented and finalized PR #41.

Key results:

• Excluded paths now remain excluded on PR and push events.
• Removed unconditional push bypasses and unused detector outputs.
• Added event/path matrix regression tests.
• Constrained Ruff to the compatible 0.8 release line.
• All 61 tests and local quality checks pass.
• Fresh GitHub CI and docs workflows pass on commit 1bd1638.
• PR is cleanly mergeable and ready for review.

---

This summary was automatically extracted from the AI working session output.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.6 Sol
• Provider: OpenAI
• Public pricing estimate: $3.414825

📊 Context and tokens usage:

• 90.0K / 200K (45%) input tokens, 11.2K / 128K (9%) output tokens

Total: (90.0K + 2.0M cached) input tokens, 11.2K output tokens, $3.414825 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.6-sol
• Thinking level: low (~7999 tokens)
• Model: GPT 5.6 Sol (gpt-5.6-sol)

📎 Log file uploaded as Gist (1184KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🎉 Auto-merged

This pull request has been automatically merged by hive-mind.

• All CI checks have passed

---

Auto-merged by hive-mind with --auto-merge flag