Mount Model Directory for Docker Container

[bennettrwu]

Transcription models are currently automatically downloaded on request by whisper-service. However, for docker containers local files are cleared when a container is run, but it is not desirable to redownload models every time the container is changed. There should be a method to mount a directory so that models are stored directly on the host machine instead.

• Add script to whisper-service to download models specified in device_configuration to some defined directory.
• Edit models to load from directory instead of automatically downloading model.
• Update README/Wiki to let user know to run script before running app (dev and prod)

Note: Depends on #12 but unrelated to #13 #14