feat: add pyperscan support

[ffontaine]

hyperscan will run simultaneously all version checkers on a file which reduce processing time.

pyperscan package is used instead of the most well-known hyperscan package as pyperscan allows to add a tag for each pattern. This feature will allow to retrieve easily the checker associated to the matched pattern.

On my local machine, running a scan on an embedded firmware takes 220 seconds with pyperscan instead of 326 seconds.

However, pyperscan is slower on a single file and unsupported on Windows, so add a --pyperscan option (disabled by default)

Fix #2485

[anthonyharrison]

Is the database built each time cve-bin-tool is run? If not, where is the database stored? Should rgere be an option to rebulid the database?

Need extra tests to support the new CLI option

[alex-ter]

A few minor docs-related comments with suggested fixes.

[alex-ter]

I've checked a bit more, and looks like pyperscan uses Vectorscan fork by default, not Hyperscan (see vlaci/pyperscan#35 and e.g., the build container configs). You might want to change that, too.

[ffontaine]

Good catch, I added a note in the README file to add this information.

[ffontaine]

Is the database built each time cve-bin-tool is run? If not, where is the database stored? Should rgere be an option to rebulid the database?

Yes, the hyperscan database is built every time cve-bin-tool is run and is not stored anywhere.
Actually, hyperscan database is handled like python regular expression which are also compiled on the fly:

cve-bin-tool/cve_bin_tool/checkers/__init__.py

Line 484 in ed4d442

cls.VERSION_PATTERNS = list(map(re.compile, cls.VERSION_PATTERNS))

As it is not saved anywhere, it doesn't make sense to add an option to rebuild it.

Need extra tests to support the new CLI option

I added a simple test, tell me if more is needed.