Renames for csv overrides PR

[pokey]

Many changes to support the new names in csv overrides PR. This PR also does the following:

• Until we properly support them, we throw an exception for subtoken exclusions. Probably not worth supporting properly until we generalize the "next" / "nth" thing in next-gen cursorless engine
• Adds a script to transform recorded test cases in bulk

Todo:

• Throw proper error for unknown action from canonicalize
• Remove old extra actions and use canonicalization instead

[pokey]

making these optional for now for backwards compatibility. Also, I went with "include" instead of "exclude" because I prefer positive booleans, otherwise it's a double negative when they're false which is a bit more cognitive load

my one hesitation with using "include" instead of "exclude" is that while these are optional, we'll be defaulting to true, which is always a bit jarring cc/ @AndreasArvidsson

[AndreasArvidsson]

I will always prefer that optional booleans default to false. That why they were defined as excluding.

[pokey]

Yes but I would prefer to make these not optional. The reason they're optional is for backwards compatibility. Hmm. I'm not seeing a way to make them backwards compatible while making them positive. Any ideas?

[AndreasArvidsson]

In Infer full targets just do.
Includeachor = target.includeanchor ?? true

[pokey]

Wait isn't that defaulting to true?

[AndreasArvidsson]

Yes and that is how it is today