[maintainability] Python-interpreter discovery is duplicated (with drift) across 3 test files

[corgab]

Summary

The same foreach (['python3', 'python'] as $candidate) { $which = shell_exec("which {$candidate} 2>/dev/null"); ... } interpreter-discovery block, followed by a skip-or-fail decision, is duplicated across three test files with small, real divergences:

• tests/Unit/Bridge/PythonBridgeTest.php:52-60 — discovers, then always markTestSkipped() if absent (no CI-fail branch)
• tests/Feature/GateParityTest.php:21-33 — discovers, then test()->fail(...) if getenv('CI') !== false, else skip
• tests/Feature/CustomProviderTest.php:22-49 (inside a named helper customProviderPythonPath()) — additionally checks getenv('AETHER_PYTHON_PATH') first, then the same discovery loop, then the same CI-fail-else-skip branch

Where

• tests/Unit/Bridge/PythonBridgeTest.php:52-60
• tests/Feature/GateParityTest.php:21-33
• tests/Feature/CustomProviderTest.php:22-49

Refactor proposal

Add one shared helper to tests/Pest.php (e.g. resolveTestPythonPath(bool $requireInCi = true): ?string), honoring AETHER_PYTHON_PATH first, then falling back to the which python3/python search, with the CI-fail-vs-skip decision as a parameter — then have all three test files call it instead of carrying their own (subtly different) copies.

Severity: low