perf(linter): compute apply_overrides rule set lazily

[shulaoda]

Config::apply_overrides is called for every linted file. It eagerly built all_rules by filtering and cloning every entry in RULES (~837 rules) into a Vec on each call.

But all_rules is only used when an override introduces a new, unconfigured plugin (to apply that plugin's category rules). In the common case — overrides that only tweak rule severities for certain file globs — it was computed and never used, cloning ~837 rules per file for nothing.

This computes all_rules lazily (Option<Vec<RuleEnum>> + get_or_insert_with), so it is only built when a new plugin actually requires it, and reused across override iterations exactly as before. Behavior is unchanged.

Both code paths are covered by existing tests: the common path (test_remove_rule, test_change_rule_severity, …) and the lazy path (test_categories_only_applied_to_new_plugins_not_in_root, test_categories_not_reapplied_to_root_plugins, …).

[codspeed-hq]

Merging this PR will not alter performance

✅ 5 untouched benchmarks
⏩ 66 skipped benchmarks¹

---

_{Comparing shulaoda:06-19-perf_apply_overrides_lazy_rules (d09e3c4) with main (b1948a1)}

Footnotes

1. 66 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩