fix: ExternalRunnerTestInfo.type matches constructor

[lf-]

I found a really funny thing in the ExternalRunnerTestInfo docs: the type field is shown as .test_type, not .type (!).

This is because it is .test_type as a field and type in the constructor. This is super confusing, let's unify the field (that I doubt anyone accesses) with the constructor so the docs make more sense.

[meta-codesync]

@facebook-github-bot has imported this pull request. If you are a Meta employee, you can view this in D104869246. (Because this pull request was imported automatically, there will not be any future comments.)

[cormacrelf]

I have definitely experienced this issue before. 👍

[JakobDegen]

(that I doubt anyone accesses)

Heh, grep in fbsource gives me 23 results.

But also, .type is a starlark builtin for getting the name, we definitely don't want to overload that. Can we go the other way instead? Add test_type as a thing we accept in the constructor, I'll volunteer to (dispath an AI to) codemod our constructors, and then we go from there? (I've also got a diff going that makes all the existing parameters named-only, that should help with that)

[lf-]

But also, .type is a starlark builtin for getting the name, we definitely don't want to overload that.

news to me! that's unfortunate, and I agree this would be a flawed approach then.

Can we go the other way instead? Add test_type as a thing we accept in the constructor, I'll volunteer to (dispath an AI to) codemod our constructors, and then we go from there?

Sounds great to me!