fix(test): strip inline tests from local-extended children of remote presets

[fohte]

Purpose

• runok test evaluates a large number of preset-authored inline tests regardless of any user override, and a user override turns those into expected allow, got deny failures
  • Example: extending github:fohte/runok-presets/base and denying find * in the user config produces a flood of failures such as find . -name "*.txt" => expected allow, got deny

Reproduction steps

1. Write the following to /tmp/test-strip.yml

   extends:
     - "github:fohte/runok-presets/base"
   rules: []

2. Running runok test -c /tmp/test-strip.yml evaluates the preset's own inline tests in bulk, e.g. 592 passed, 0 failed, 592 total

Approach

• Once an extends chain crosses into any remote preset, treat every descendant as preset-authored and strip its inline tests and top-level tests: block
  • The strip introduced in #227 decided based on the loaded reference's own kind, so a remote preset like runok-presets/base whose extends: [./readonly-unix.yml, ...] reaches its siblings via local paths left those siblings outside the strip
• Tests in the user's own config and in presets the user extends directly via a local path are preserved as before

Design decisions

Where to apply the strip

Decision	Design	Pros	Cons
Chosen	Keep the existing strip in load_preset_with and additionally strip in resolve_extends_recursive while propagating an under_remote flag down the chain	Direct callers of load_preset that skip extends resolution (such as the version check in update-presets) keep stripping the remote root unchanged	The strip lives in two places, so future strip targets must be added through strip_preset_tests so both call sites stay in sync
Rejected	Move the strip out of load_preset_with and centralise it in resolve_extends_recursive	The strip has a single responsibility	Direct callers of load_preset that skip extends resolution would stop stripping the remote root, changing existing behaviour

[codecov]

Codecov Report

❌ Patch coverage is 95.74468% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.89%. Comparing base (1feddf1) to head (c629873).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/config/preset.rs	95.74%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #339      +/-   ##
==========================================
+ Coverage   88.78%   88.89%   +0.11%     
==========================================
  Files          53       53              
  Lines       12322    12369      +47     
==========================================
+ Hits        10940    10996      +56     
+ Misses       1382     1373       -9     

Flag	Coverage Δ
Linux	88.72% <95.74%> (+0.13%)	⬆️
macOS	90.04% <95.74%> (+0.10%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gemini-code-assist]

Code Review

This pull request fixes a bug where runok test would evaluate inline tests from presets reached via a remote ancestor, even if those presets were referenced through local paths. The fix involves tracking the remote status through the recursive extension resolution and stripping tests from any preset loaded under a remote boundary. A review comment correctly identifies that the new tests repeat setup logic for PresetCache and CacheMetadata, which should be extracted into a #[fixture] according to the repository's style guide to improve maintainability.