Major refactor of PHP code + improvements and bug fixes

[chrisjp]

The little proxy.php file had become a big mess of if/else statements and a lot of repeated code. This refactoring takes a more OOP stance and makes it significantly easier to add new TTS APIs. Additionally, some improvements and bug fixes are included.

New:

• Added StreamElements and made it the default service as it has a lot more voices compared to Streamlabs
• More TikTok voices added
• Hundreds of Microsoft Azure Speech voices added (via Bing Translator)
• Docker compose file included for quickly running your own local copy of the demo site
• Basic stats page (won't be particularly interesting unless you're hosting a very public website)

Changes:

• Renamed Polly to Streamlabs to more accurately reflect the API being used, and because StreamElements also makes use of Polly voices which could cause confusion
• Redesigned playlist page to show full transcript and highlight the text currently being spoken
  • also fixes the bug where line breaks weren't respected
• Automatically add a dialogue box for text entry when at least 1 voice is selected on conversation page
• JavaScript now docblocked. Voices JSON and helper functions now split into their own respective files
• If saving transcripts alongside audio files it will now be saved as JSON rather than plain text

Bug fixes:

• undefined or otherwise erroneous URL parameters are now handled and fallback to a default if necessary
• Generated audio URLs will now be served over http if you don't have https set up
  • I really hope you do have https enabled - this was more for when running a local only instance
• Attempts to create AUDIO_DIR if it does not exist.
• Failure to save audio locally will not fatally error and instead return the remote URL