Finish skeleton common-accessioning robot and workflow def for speechToTextWF

[jmartin-sul]

The start of the code for this is already there, and assumes the name will be captionWF: https://github.com/sul-dlss/common-accessioning/tree/main/lib/robots/dor_repo/caption

But part of this work will be renaming in light of the decision to go with speechToTextWF. See terminology note on sul-dlss/speech-to-text#1

The workflow XML is not yet in place, but the skeleton code is already present.

The workflow definition would be a new XML file here: https://github.com/sul-dlss/workflow-server-rails/tree/main/config/workflows

The skeleton workflow with placeholders for steps we expect, with implementations filled in as supporting services are developed, will be similar to what we did for ocrWF.

Steps we're likely to have, in order:

• writing media files to S3
• (maybe) signaling that those files need to be transcribed (if simply writing them to the bucket isn't enough)
• picking up a workflow step signaling that the transcription output has been placed in an S3 bucket
• updating the cocina appropriately and accessioning the output files

Currently, there is placeholder workflow step code for starting the workflow, ending it, and generating speech-to-text output (this last part will very likely be broken into the multiple steps describe above).

[peetucket]

Closed by #1342