Give transcript runner an option to generate screen recordings

If you look at: https://www.unisonweb.org/docs/quickstart, there's some asciinema recordings that were painstakingly recorded by hand and need to be updated by hand every time there's changes to UCM. It would be great if we could automate this in some way.

Our thought was to give the transcript runner the ability to produce asciinema files directly, and this would be super cool and a nice self-contained little project. (The transcript runner stuff is in TranscriptParser.hs.)

As we thought more about it though, we ended up with more questions than clarity:

• The asciinema files in the quickstart guide maybe don't lend themselves to transcripts.

  • The first demonstrates bash commands, which is separate from the transcript runner.
  • The second demonstrates git output, which also isn't currently captured by the transcript runner.
  • The third shows interleaved ucm and editor work, which seems cool, but we're not sure how to structure it as a transcript (see next bullet).

• Should asciinema be produced from a single code block e.g. ucm:asciinema or out of a full transcript and its multiple code blocks? Note again that interleaving editor and ucm work involves multiple ucm and unison code blocks, but for an arbitrary transcript, it might not make sense to run all the steps in a video at the top of the file.

• Is there a restricted use case that this would help, even if it's not the general case?