Config option to make resolvers return values instead of pointers

[ianling]

#1393

Adds a config option to make resolvers return values instead of pointers.

For example, given the following schema:

type User {
  id: ID!
  name: String!
}

type Query {
  user: User!
  maybeUser: User
}

Old behavior before this PR (matches default behavior after this PR)

func (r *queryResolver) User(ctx context.Context) (*User, error) {
	panic("not implemented")
}

func (r *queryResolver) MaybeUser(ctx context.Context) (*User, error) {
	panic("not implemented")
}

New behavior with resolvers_always_return_pointers: false (*User vs User in return type of first function):

func (r *queryResolver) User(ctx context.Context) (User, error) {
	panic("not implemented")
}

func (r *queryResolver) MaybeUser(ctx context.Context) (*User, error) {
	panic("not implemented")
}

I have:

• Added tests covering the bug / feature (see testing)
• Updated any relevant documentation (see docs)

[ianling]

Holding until my other PR is finished/merged

[coveralls]

Coverage decreased (-0.05%) to 75.013% when pulling 770d6eb on ianling:resolvers_not_always_pointers into ddd825e on 99designs:master.

[nikitacrit]

@ianling Thanks, I'm really looking forward to it. Any predictions when it will be merged?

[StevenACoffman]

Sorry! I wanted to let the release sit for a few days for any urgent fixes for the recent WebSocket changes (since we don't use those in production at Khan Academy, so I always worry about breaking someone's edge case). I merged the other PR though.

[ianling]

Fixed merge conflicts.

Don't be afraid of the large change size, the actual change is only like 1 line, the rest is auto-generated models for tests.

And just to clarify, the default behavior of gqlgen remains the same after this PR is merged; users of gqlgen will only see different behavior if they add the new config option to their config.

[ianling]

Not sure why the Windows tests failed.