cpe: unbind fs optimizations

[hdonnay]

Some memory profiling revealed that in large imports, this function is doing a lot of allocation just to do the unbinding (meaning the underlying cpe.WFN value is not in that bookkeeping).

These changes include:

• some cleanup
• benchmarking
• a new, generated parser
• switching to the new parser

The results locally:

goos: linux
goarch: amd64
pkg: github.com/quay/claircore/toolkit/types/cpe
cpu: 11th Gen Intel(R) Core(TM) i7-11850H @ 2.50GHz
BenchmarkUnbindFS/Simple-16         	 1978447	       655.0 ns/op	      32 B/op	       3 allocs/op
BenchmarkUnbindFS/Escape-16         	 1889960	       641.0 ns/op	      24 B/op	       2 allocs/op
BenchmarkUnmarshalFS/Simple-16      	 3437407	       361.9 ns/op	       0 B/op	       0 allocs/op
BenchmarkUnmarshalFS/Escape-16      	 2744743	       455.2 ns/op	      16 B/op	       1 allocs/op
PASS
ok  	github.com/quay/claircore/toolkit/types/cpe	6.532s