[WASM ABI 1.0] Change ColId from u32 to u16

[Centril]

Description of Changes

• Changes the definition of ColId to be u16, as per https://github.com/clockworklabs/SpacetimeDBPrivate/pull/863.
• Introduces GroundSpacetimeType for the {Identity, ColId, ...}::get_type cases to ensure that cases don't get missed.
• Gets rid of create_index, , as per https://github.com/clockworklabs/SpacetimeDBPrivate/pull/863. There were no users in the high level API.

cc #1460

API and ABI breaking changes

The actual ABI signature doesn't change here as WASM represents u16 as u32, but the behavior does change.

Expected complexity level and risk

2

Testing

• ColList tests were run through miri.

[kazimuth]

All makes sense.

I guess I should change names in my other PR to ABI v10? Or maybe we should detach the RawModuleDef version from the WASM ABI version.

[kazimuth]

I think it's slightly stronger than this.
Because e.g. (x: (y: u8) | (z: u16)) satisfies this condition, but I believe should not be considered a "ground type". This is because AlgebraicType contexts is very flexible -- you can either directly include a SumType in a ProductType, or you can include it behind a Ref.

AlgebraicTypeUseForGenerate that we discussed in #1590 will resolve this since it will only allow Sum or Product types behind Ref. Once that's defined, we can say "can be represented as AlgebraicTypeUseForGenerate without any Refs." (man, that's a mouthful...)

[Centril]

Formally the type you mentioned is a ground type though. It has no free variables, which is the definition of a "ground term" / expression.

[kazimuth]

Ah, I did not know that was a standard term :) I thought it just meant "builtin".

[Centril]

Or maybe we should detach the RawModuleDef version from the WASM ABI version.

Yeah probably, for now, and in any case, I'd like to lower the ABI to spacetimedb_1 just before we launch mainnet.