Perf: cache UseStatementSniff getUseStatements across fix iterations

[dereuromark]

Summary

Adds a per-file cache to UseStatementSniff::getUseStatements(), the same shape as #64 and #65.

UseStatementSniff has its own getUseStatements() implementation that duplicates UseStatementsTrait::getUseStatements() (the one cached in #64). The existing instance-level cache via $existingStatements covers repeated calls within a single phpcs pass; the new static cache adds coverage across phpcbf fix iterations, where populateTokenListeners() creates fresh sniff instances per pass and resets the instance cache.

Cache invalidation

Same fingerprint-based scheme introduced in #64. Token count alone is not strong enough — an alias rename keeps it constant. Cached entries record a content fingerprint of each use statement range and re-verify them against the live tokens before being trusted.

Defensive belt-and-braces: refuse to serve a cached empty result. Files with no use statements don't benefit from caching anyway, and skipping the cache for empty results means we won't serve stale state in the corner case where a fix adds a first use statement while another simultaneous fix happens to keep the overall token count unchanged.

Why not the trait

UseStatementSniff::getUseStatements() is almost a duplicate of UseStatementsTrait::getUseStatements() (the one cached in #64), but with a slightly different result shape (no 'statement' field). Refactoring this sniff onto the trait is doable but out of scope for a perf-only PR — same cache shape, kept local.

Benchmarks

Measured on the same 1095-file CakePHP 5 app from #62 / #63 / #64 / #65. Per-sniff CPU times from phpcs --parallel=1 --report=performance:

Sniff	Before this PR	After this PR
PhpCollective.Namespaces.UseStatement	~3.36s	~2.08s (-38%)

Note

An earlier draft of this PR also added a cache to FullyQualifiedClassNameInDocBlockSniff::parseUseStatements() / ::getNamespace(). Those changes were dropped after #65 landed with a different (monotonic-stack-pointer) caching scheme covering the same hot path.

Verification

• Existing PHPUnit suite passes (100 tests / 122 assertions).
• composer cs-check and composer stan on this repo: clean.
• Output diff on the activity-pro test corpus: identical violations before vs after.
• codex review --uncommitted on the original (broader) draft flagged P2 cache-invalidation concerns; the empty-result skip in this final version was added in response.