Generate air.schema.json using schemars

[DavisVaughan]

First step of #106

This PR:

• Adds a "schemars" feature to the settings and workspace crates that optionally derive JsonSchema. Note that schemars is smart enough to respect serde attributes, like rename_all = "kebab-case" and deny_unknown_fields, which is awesome
• Generates the air.schema.json using a new xtask helper, callable through just gen-schema
  • There are tests here to ensure that we can't accidentally allow our schema to get out of sync with our TomlOptions struct. If the options struct is updated, the schema must also be updated.

This is a first step, we also need to either:

• Do a PR to add our schema to https://github.com/SchemaStore/schemastore. We would be in charge of keeping it up to date when we make changes to the schema (we would just do another PR and add this to our release checklist). The Even Better Toml vs code extension knows how to look there and download the air schema on the fly. Putting it here also enables other IDEs to gain access to our schema, for example, Zed has a TOML extension that uses taplo (the even better toml lsp) as well, so it will also be able to access the air schema
  • Update: We can also set our schema as self hosted, so we never have to do another PR ever again to schemastore https://github.com/SchemaStore/schemastore/blob/master/CONTRIBUTING.md#how-to-add-a-json-schema-thats-self-hostedremoteexternal
• Alternatively, add the schema in our VS Code extension as a tomlValidation contribution point, which Even Better Toml will look for and load automatically https://taplo.tamasfe.dev/configuration/developing-schemas.html#visual-studio-code-extensions. This will not benefit other IDEs than VS Code and Positron. It has the small benefit of not needing to PR into schemastore when we make a change, but I'm hoping that won't matter much, so I don't think we should do this approach.

To immediately try this out, you can put this in your settings, then make sure to run Developer: Restart Extension Host

    "evenBetterToml.schema.associations": {
        "^.*air.toml$": "https://raw.githubusercontent.com/posit-dev/air/8049e67080e0976bfe8bcf60efa30912a88ddea1/air.schema.json"
    },

Then open an air.toml and it should do this

Screen.Recording.2025-02-07.at.1.20.54.PM.mov

As you can see in the video, there are some issues with field duplication. I do not believe this is our fault, I think this is this taplo issue tamasfe/taplo#514. Ruff also has the same issues. I do not believe we should let this block us, as this is still quite useful even with the duplication.