I don’t think that’s the actual problem. GitHub does not change the encoding of files in Git repos. Nor (generally speaking) does Git itself [0]
It appears most of the encoding issues were introduced by manually editing files, in response to a corporate policy that certain comments be censored (mainly comments that could be viewed as unprofessional). While it is possibly to do this without altering the encoding of the files, it appears the person who did it used an editor that rewrote the file encoding, and didn’t realise it was doing that
[0] Git can be configured to do automatic conversion between bare LF and CRLF newline formats. That’s generally a bad idea. It is mainly people on Windows who turn it on, although nowadays many Windows tools can handle Unix style line endings without issue, making it largely unnecessary. Also, you can install Git repository hooks that alter file contents as they are being committed, although the majority of repos don’t have any such hooks installed