`speechToTextWF` `update-cocina`: update cocina structural to reference files generated by speech_to_text_generation_service

[jmartin-sul]

for now, blocked by the implementation of preceding steps in speechToTextWF, but we could probably parallelize this with work on the preceding steps, if we wanted. we might be able to work off of the ocrWF equivalent as an example?

this is the speechToTextWF equivalent of ocrWF update-cocina

stub code in the speechToTextWF: https://github.com/sul-dlss/common-accessioning/blob/main/lib/robots/dor_repo/speech_to_text/update_cocina.rb

part of https://github.com/sul-dlss/common-accessioning/issues/1363

[peetucket]

Note: at the moment, this step is implemented using the exact same algorithm used by the ocr update-cocina step, which means it picks up new files in the workspace generated by Whisper, finds resources with base filenames that match, and then adds them to the resource.

We still need to consider:

1. which files get roles (e.g. transcription)
2. if all files should be added or only of a certain mimetype

[jmartin-sul]

the developer who picks this up should consult with andrew on whether the code we have is doing the right thing, or if it needs tweaking

[peetucket]

Roles:

".vtt" = "caption" ".txt" = transcription

Publish/preserve/shelve = true for vtt and txt files

[jmartin-sul]

leaving open till we confirm that the captions generated by speechToTextWF actually work in sul-embed when watching a video

[jmartin-sul]

captions confirmed to display correctly

[peetucket]

Here is one to check: https://sul-purl-stage.stanford.edu/bh691ds2057

I see a transcript panel in sul-embed