OpenAPI validations fails due to incorrect type of "dimension_types"

[ml-evs]

This is the cause of Materials-Consortia/OPTIMADE#314, which @CasperWA has summarised below.

The problem is that we are using (approximately) Tuple[PERIODICITY, PERIODICITY, PERIODICITY] when constructing dimension_types, but the pydantic docs (link below) states that:

Python type	JSON Schema Type	Additional JSON Schema	Defined in
Tuple[str, int]	array	{"items": [{"type": "string"}, {"type": "integer"}]}	JSON Schema Validation

And equivalently for any other set of subtypes. Note: If using schemas for OpenAPI, you shouldn't use this declaration, as it would not be valid in OpenAPI (although it is valid in JSON Schema).

Steps to take for the fix:

• Run an openapi validator in our CI.
• Switch away from Tuple throughout our API models (what should we use instead?)

From @CasperWA:

I notice that we still don't validate properly according to swagger (see recent issue on the spec repo)...
https://validator.swagger.io/validator/debug?url=https://raw.githubusercontent.com/Materials-Consortia/optimade-python-tools/a6926f53263fcabc91523e974c4d4b5bbe5bc9f1/openapi/openapi.json

Okay, I think I may have found the issue here.
According to the OpenAPI specification the items property of a Schema Object MUST be an object. But for dimension_types, it's an array.
Edit: The issue is also mentioned in pydantic's documentation here (see in the table for Tuple).

Originally posted by @CasperWA in #477 (comment)

[CasperWA]

• Run an openapi validator in our CI.

We already are (openapi-spec-validator).
But perhaps we should switch to requests to https://validator.swagger.io/validator/debug?url=https://raw.githubusercontent.com/Materials-Consortia/optimade-python-tools/tree/master/openapi/openapi.json ?

• Switch away from Tuple throughout our API models (what should we use instead?)

I have done this by using conlist.