refactor: use ignore crate for matching gitignore patterns

[beeb]

Closes #337 and #2

The benchmark comparing ambs 0.6 with this branch:

$ hyperfine --warmup 3 "./target/release/ambs EXPORT_SYMBOL_GPL ./data/linux" "ambs EXPORT_SYMBOL_GPL ./data/linux"
Benchmark 1: ./target/release/ambs EXPORT_SYMBOL_GPL ./data/linux
  Time (mean ± σ):      1.217 s ±  0.046 s    [User: 2.356 s, System: 2.507 s]
  Range (min … max):    1.173 s …  1.309 s    10 runs

Benchmark 2: ambs EXPORT_SYMBOL_GPL ./data/linux
  Time (mean ± σ):      1.191 s ±  0.040 s    [User: 2.283 s, System: 2.529 s]
  Range (min … max):    1.131 s …  1.242 s    10 runs

Summary
  ambs EXPORT_SYMBOL_GPL ./data/linux ran
    1.02 ± 0.05 times faster than ./target/release/ambs EXPORT_SYMBOL_GPL ./data/linux

$ hyperfine --warmup 3 "./target/release/ambs irq_bypass_register_producer ./data/linux" "ambs irq_bypass_register_producer ./data/linux"
Benchmark 1: ./target/release/ambs irq_bypass_register_producer ./data/linux
  Time (mean ± σ):      1.209 s ±  0.023 s    [User: 2.220 s, System: 2.499 s]
  Range (min … max):    1.167 s …  1.256 s    10 runs

Benchmark 2: ambs irq_bypass_register_producer ./data/linux
  Time (mean ± σ):      1.177 s ±  0.016 s    [User: 2.123 s, System: 2.525 s]
  Range (min … max):    1.157 s …  1.214 s    10 runs

Summary
  ambs irq_bypass_register_producer ./data/linux ran
    1.03 ± 0.02 times faster than ./target/release/ambs irq_bypass_register_producer ./data/linux

I think it's a reasonable regression in performance considering the new implementation respects the full scope of gitignore syntax.

The ignore crate also provides Walk which can perform the task of traversing the file tree while respecting the gitignores. It is based on walkdir internally. That might provide additional performance benefits if implemented in the future.

[dalance]

Thank you for your contribution!
I agree the performance regression is reasonable.
I'll merge this PR.