chore(SDL): refactor _legacy.graphql into individual typeDefs

[mattkrick]

Description

Refactoring into the SDL pattern has been a pain because we need to remove the typeDef from the _legacy.graphql file & put it into its own typeDef. We do this because it's 8000 LOC & ripe for merge conflicts.

I wrote some code that did this for us for every type.
Now refactoring to the SDL pattern just requires transforming the resolve statements, no need to touch .graphql files 🎉

How is it broken out?

• Every type has its own file, even one-liners like unions & scalars
• Mutations & Queries are types too, so you'll see a Mutation.graphql
• Additionally, we have _xGitHub and _xGitLab.graphql, which override their integrations

[github-actions]

This PR exceeds the recommended size of 1000 lines. Please make sure you are NOT addressing multiple issues with one PR. Note this PR will be delayed and might be rejected due to its size.

[Dschoordsch]

I liked the previous way it was split into multiple files better: put each mutation in its own file and add the related return and input types there too. The second part of this might be a bit harder to automate, but then you have all related types in one file and don't need to jump between them too often.

[mattkrick]

It was definitely nice grouping them together & having fewer files!
The question I kept asking myself is, "why are we treating the Mutation object differently from other objects?"

If the mutation returned a User object, we wouldn't put the User definition in there, we'd give it its own file because it's shared across definitions.
Using that same logic, UpdateMeetingPromptSuccess type is shared with the Subscription object, so why does that get grouped with in the same file as the mutation field?

Moving forward, I'm down to keep using out extend pattern! TBH I don't really care where the defs are stored, I just use search anyways. I just wanted to get rid of the stale definitions in _legacy.graphql that were holding us back.

How should we proceed? Could we merge this now, or should we refactor manually?

[Dschoordsch]

It's just a preference. I don't see a reason why we shouldn't merge it as is. Better than manually moving types out of _legacy.graphql.

[nickoferrall]

Love this!