Add performance benchmark suite and scan profiling output

[ColumbusLabs]

Goal

Keep DebtLens fast and predictable as more rules and project shapes are added.

Difficulty

Medium

Area

Performance

Suggested approach

• Add benchmark fixtures for small, medium, and large synthetic projects.
• Track total scan time and per-rule timing in an opt-in mode such as --profile.
• Warn when a detector hits configured caps or dominates runtime.
• Document expected performance budgets and how to run benchmarks locally.

Acceptance criteria

• --profile reports per-rule timing without changing findings.
• Benchmarks are deterministic enough for local comparison.
• CI can run a lightweight benchmark smoke without being flaky.

Suggested test command

npm test -- tests/core/scan.test.ts tests/cli/scan.test.ts

[ColumbusLabs]

Working on this — PR incoming.

[ColumbusLabs]

Benchmark suite PR incoming. Per-rule profiling remains in open #62.