codegen: fix required ID deserialization

[ajwerner]

Fixes #522

[tomhoule]

Good catch. Thanks for the fix!

[danieleades]

this was merged with failing CI jobs, will need to be fixed in a new PR to get the builds passing again

[Sytten]

This broke a bunch of stuff and needs to be reverted or improved:

`?` operator cannot convert from `std::string::String` to `std::vec::Vec<std::string::String>`
expected struct `std::vec::Vec<std::string::String>`
   found struct `std::string::String`

= note: `?` operator cannot convert from `std::string::String` to `std::option::Option<std::vec::Vec<std::string::String>>`
   = note: expected enum `std::option::Option<std::vec::Vec<std::string::String>>`
            found struct `std::string::String`

[emilbonnek]

@tomhoule,
I’ve just upgraded to 0.15.0, and I’m running into what looks like a similar issue to the above:

error[E0308]: `?` operator has incompatible types
 --> myfile.rs:8:10
  |
8 | #[derive(GraphQLQuery)]
  |          ^^^^^^^^^^^^ expected `Option<Vec<String>>`, found `String`
  |
  = note: `?` operator cannot convert from `std::string::String` to `std::option::Option<Vec<std::string::String>>`
  = note: expected enum `std::option::Option<Vec<std::string::String>>`
           found struct `std::string::String`
  = note: this error originates in the derive macro `GraphQLQuery` (in Nightly builds, run with -Z macro-backtrace for more info)

My query looks like this:

mutation FileDelete($input: [ID!]!) {
    fileDelete(fileIds: $input) {
        deletedFileIds
        userErrors {
            field
            message
        }
    }
}

and my macro like this:

#[derive(GraphQLQuery)]
#[graphql(
  schema_path = "schema.json",
  query_path = "myfile.graphql",
  variables_derives = "Debug, Clone",
  response_derives = "Debug, Clone",
  skip_serializing_none
)]

I’ve been digging into it for a while, but I haven’t been able to find a workaround.
Do you know if the concern raised earlier by @Sytten was addressed before the release, or if this might have been an unintended regression? It’s also possible there’s just a breaking change I haven't understood. Any guidance would be appreciated.