adding whisper asr API support

[GRbit]

Hi @FlyingFathead !

Thank you again for your project. I'm currently running some voice transcription on my server and it's a pity that we have so many useful open-source service that can do voice transcription, but every one of them starts it's own whisper model. I just have no memory for all the things I want.

So, that where this PR comes from, it enables this bot to use whisper-asr-webserive. API mode requires aiohttp, which is included in requirements.txt. There are no other dependencies.

My IDE also applied auto-formatting for all the spaces left at the end of the lines, hope it's ok. Really new code is in whisper_api_client.py and in the new functions transcribe_audio_api (and transcribe_audio_local, but it's more of a refactoring).

I'll be honest with you, I used LLMs to generate most of the code, but I read it and I tested it. This version with remote transcription enabled is currently working on my server.

If you like the idea, I have some plans to continue with adding translation and maybe something else.

[FlyingFathead]

Thanks for the PR -- interesting idea, but I am not comfortable merging it as-is yet.

A few things need tightening first:

• In API mode, the selected model appears to be ignored entirely by the client.
• The same audio file gets uploaded 3 times per job (txt, srt, vtt), which is... probably unintentional?
• The client reads the whole file into memory before upload.
• Exception handling is too broad right now and may hide actual bugs behind API fallback/failure behavior.
• The README currently overstates compatibility, it looks like this looks more like support for a specific /asr API shape than a generic Whisper-compatible API.

If you want to revise those parts, I’m happy to take another look. Cheers.

[FlyingFathead]

Also, one maintainer-side concern here: I do not use this setup myself and have not tested it, so merging it would mean taking on a feature path I am not really in a good position to maintain or debug over time. The idea itself makes sense, but I need to be careful about merging integrations I do not actively use and am not familiar enough with to support properly.