Syntactic tweaks to models and schemas for compatibility with fastapi>0.66

[ml-evs]

This PR makes 2 main changes to ensure FastAPI compatibility going forward (see #887).

The issue is essentially that FastAPI broke OpenAPI specs generated from pydantic by adding extra = 'allow' to its base OpenAPI schema class (see tiangolo/fastapi#3745). For example, if you used Field(const=True) in pydantic, the const key would be added to the eventual OpenAPI schema, yet const does not exist as an OpenAPI field. The same then happened for our custom schema injections of support, queryable and unit.

This PR adds the following workarounds:

• Replace all uses of const and pattern with just the pydantic regex (which checks the value with the regex and also injects pattern into the schema). typing.Literal, replacing constant values in the OpenAPI schema with single-valued enums. The other option here would be to remove the const check and replace pattern with regex so that pydantic applies the regex during validation.
• Add wrappers to the underlying pydantic fields that prefix our custom injected fields, so that support becomes e.g. x-optimade-support. These need to be retained in the JSONSchema so that they can be used to set the OpenAPI nullable flag.
• Finally this PR fixes an issue where one field was not wrapped in OptimadeField already.

[ml-evs]

Thanks for the review @JPBergsma, I've put this PR on hold as I think this is going to more complicated than I first thought.

[CasperWA]

Concerning const, did you see that the value MUST be equal to the default value? https://pydantic-docs.helpmanual.io/usage/schema/#field-customization

[CasperWA]

Concerning const, did you see that the value MUST be equal to the default value? https://pydantic-docs.helpmanual.io/usage/schema/#field-customization

Ah, I see this was already the case. Never mind :)

[ml-evs]

Just FYI, I got a response on my FastAPI issue, it seems like there will not be a fix.

They accepted my PR adding a warning to the docs: https://fastapi.tiangolo.com/tutorial/body-fields/#add-extra-information

[ml-evs]

Thinking about this more, we might just have to bite the bullet and change how we test the OpenAPI schema. It looks like MP can't update to the latest OPTIMADE version atm as they are already using a more recent FastAPI. I'll see what I can do in time for the v0.18 release.

[codecov]

Codecov Report

Merging #1131 (138fddb) into master (8d1edba) will decrease coverage by 0.11%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1131      +/-   ##
==========================================
- Coverage   91.95%   91.84%   -0.12%     
==========================================
  Files          72       72              
  Lines        4265     4279      +14     
==========================================
+ Hits         3922     3930       +8     
- Misses        343      349       +6     

Flag	Coverage Δ
project	91.84% <100.00%> (-0.12%)	⬇️
validator	91.23% <100.00%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
optimade/models/links.py	100.00% <ø> (ø)
optimade/models/optimade_json.py	95.78% <ø> (ø)
optimade/models/references.py	97.72% <ø> (ø)
optimade/models/structures.py	95.75% <ø> (ø)
optimade/server/mappers/entries.py	97.11% <ø> (ø)
optimade/server/routers/index_info.py	100.00% <ø> (ø)
optimade/server/routers/info.py	95.45% <ø> (ø)
optimade/models/baseinfo.py	95.23% <100.00%> (ø)
optimade/models/entries.py	97.77% <100.00%> (ø)
optimade/models/index_metadb.py	100.00% <100.00%> (ø)
... and 3 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8d1edba...138fddb. Read the comment docs.

[JPBergsma]

You have listed "x-optimade-unit" unit twice here. That seems a bit strange.

[ml-evs]

Hmm yes, this is a typo, though not a consequential one. I will remember to fix it in the next PR I make