[speech-to-text]: problems with local development

[cabljac]

It is very difficult to work on this extension locally, because it relies on the production speech to text api (which uses a long running job/operation, and writes back to a storage bucket itself)

We could write some mock transcription function just for local development, though this also has some issues/caveats:

• There seems to be a mismatch between the file system of firebase functions and the local environment

• This would also mean we have to maintain this mock API (The production one may change)

Open to other solutions.