Structural improvements

[mhubii]

Issues:

• Voice activation threshold not met well
• Period buffer as opposed to listening buffer
• Namespaces and launch files
• [x] Move implementation to ros-ai?

Suggestion for v1.0.0:

• [x] Utilize whisper.cpp
  • [x] Create a vendor package for it
  • [x] Run locally
• [x] Run whisper inference in cpp
  • [x] Switch from ament_python to ament_cmake
  • [x] Compile against vendor
• [x] Keep audio listener in python (server style via trigget requests)
• [x] Create launch mixins
• [ ] Package and upload to ros-index

[mhubii]

done, except for ros-index