Tools: Port refraction's bulk compression script to Python

[TheTechnician27]

Description of Changes

Converts refraction's bulk compression script which was presented as a Windows-only batch file into a cross-platform Python script.

Rationale behind Changes

Makes the tool cross-platform, putting Windows, Linux, and macOS users (freeBSD?) on a more equal footing and allowing us to ship the tool with PCSX2 instead of tucked away in Red's PCloud.

Suggested Testing Steps

Try all 9 options and try them with and without deleting source items as well as different block sizes if compressing. Presently needs tested on all three major OSes, and it may work in freeBSD, but there's no guarantee at all for that. Until it has been properly tested by multiple people, USE THIS IN A CONTROLLED TESTING ENVIRONMENT WITH JUST A FEW FILES WHICH YOU HAVE BACKED UP ELSEWHERE.

[fizwidget]

I’ve been using createdvd with my games. Just wondering why createraw is used here - is there a benefit, or much of a muchness?

[refractionpcsx2]

I made the script before createdvd was an option, but also I find createraw tends to give slightly better compression, from my own tests.

[RedDevilus]

Also they share similar options, createraw has more versatility in the way it has more options like createdvd doesn't have No --unitsize Option:'

Both have the same compression types (default is lzma,zlib,huff,flac).

But with createdvd you can easier give the correct hunksize. Createraw will default to correct one but you can change to whatever for more or less compression.

Generally looking at the options they do have the same compression powers just differently handled with the code logic.

[crashGG]

Is there any way to modify the compression judgment condition of chdman? According to my observation, when chdman currently selects multiple compression algorithms for mixed compression, the final compression algorithm is only roughly judged based on the final block size.This will give the smallest final chd size, but it's not smart enough.
If there is ,for example, when the compressed block size exceeds 95% of the original block, the compression is abandoned and the block is stored directly. When the lzma compressed block size is less than 90% of the zstd compressed block, it is used, otherwise the zstd compressed block is still used. This won't be the smallest size, but it's the most balanced approach, giving you a good balance between size and cost. Maybe.