fix: premarshal structs get generated with omitempty tag

[kevinmichaelchen]

This PR addresses a bug described in https://github.com/Khan/genqlient/issues/263

Basically, whenever a struct involves a custom genqlient binding, a secondary "premarshal" struct gets generated.

The bug was that this "premarshal" struct was not propagating the omitempty JSON tags, which was resulting in unexpected behavior.

The fix involved a few changed lines in the Go template, and a few changes in the unit tests.

I have:

• [x] Written a clear PR title and description (above)
• [x] Signed the Khan Academy CLA
• [x] Added tests covering my changes, if applicable
• [x] Included a link to the issue fixed, if applicable
• [ ] Included documentation, for new features
• [x] Added an entry to the changelog

[benjaminjkraft]

Thanks! I'll take a closer look later but a few quick comments now:

• Since this is a bugfix it probably doesn't need any documentation changes, just a brief changelog entry summarizing and linking to the issue.
• If it's not too hard, it would be nice to write an integration test that fails without this change and passes with it! Not sure offhand how hard that will be so if it's too hard it's probably ok to skip.
• Similarly if it's not too hard you could add another unit test case with a schema that looks more like the hasura one. But I'm not sure how interesting that would be, maybe there's no need since existing tests are hitting this case.

[kevinmichaelchen]

a schema that looks more like the hasura one

Here's what I'm thinking for that:

Click me

```graphql ################## # Hasura.graphql # ################## query GetPokemon($where: getPokemonBoolExp!) { getPokemon(where: $where) { level } } ################## # schema.graphql # ################## type Query { getPokemon(where: getPokemonBoolExp): [Pokemon!]! } input getPokemonBoolExp { _and: [getPokemonBoolExp!] _not: getPokemonBoolExp _or: [getPokemonBoolExp!] level: IntComparisonExp } input IntComparisonExp { _eq: Int _gt: Int _gte: Int _in: [Int!] _isNull: Boolean _lt: Int _lte: Int _neq: Int _nin: [Int!] } ```

[kevinmichaelchen]

OK, I managed to get snapshot generation working with

UPDATE_SNAPSHOTS=1 go test ./...

The challenge I'm now facing — and maybe this is not relevant to the original ticket — is to get omitempty tags on the fields inside IntComparisonExp

[benjaminjkraft]

Great, thanks! I think IntComparisonExp is another limitation, being discussed in #260 (there are a few ways around but none quite convenient for these hasura schemas). Anyway it's good to have the test; I think this weekend I'm going to try to understand what "make genqlient and hasura not fight so much" might mean more generally.