[Draft] Skip null values when serializing in rust

[mssalemi]

Addresses: https://github.com/shop/issues-shopifyvm/issues/318

Summary

• Rust functions were outputting null fields like "attributes": null, "image": null, "price": null in JSON responses, making them larger than necessary

What this PR does?

• Skips null fields: Optional fields with None values are no longer included in the serialized output
• Keeps required fields: Non-optional fields are always serialized as before
• Dynamic field counting: Adjusts msgpack field count based on actual fields being serialized

Before:

  {
    "attributes": null,
    "image": null,
    "price": null,
    "title": "Something that is wrapped"
  }

After:

  {
    "title": "Something that is wrapped"
  }

Approach

• Analyzes GraphQL schema at compile time to detect nullable vs required fields
• conditionally serialization for optional fields (if Some(value) { serialize })
• Two-pass msgpack serialization (count fields, then serialize) to maintain format

Testing

• All existing tests pass (no breaking changes)
• New test added to verify null field skipping behavior
• Backwards compatible with existing function implementations

Pro's

• Smaller JSON output size
• Cleaner, more predictable function responses

[mssalemi]

Not super familiar with bluejay, but I think if we have something like:

ivd.r#type().is_nullable() 

Then we could avoid using this approach. We need to get the type information so we can match correctly. I ran into issues when trying to use matching because some types were required (aka not optional), and some were optional.

[adampetro]

Yeah we should be using bluejay here for the type information

[adampetro]

Can we just consolidate this into one line where we pre-compute the number of non-nullable fields and add the rest based on value, e.g. we might end up with:

let field_count: usize = 5 + self.foo.is_some().into() + self.bar.is_some().into();