fix: preserve TTS audio content type

[ivan-digital]

Summary

• detect WAV, Ogg, FLAC, M4A, WebM, and MP3 containers from TTS response bytes
• return the matching Content-Type from workspace TTS responses
• send Telegram audio with the matching MIME type and filename extension
• retain MP3 as the fallback for unknown data to preserve existing behavior

Testing

• npx --yes jest@29.7.0 server/__tests__/utils/TextToSpeech/audioFormat.test.js --runInBand
• Prettier check for all changed files
• Node syntax checks for changed runtime files

Fixes #6011

[timothycarambat]

@ivan-digital this is beautiful and a very welcome change. Thanks for taking the time to make this + following contrib guidelines. Personally appreciate the PR.