[RyuJit/WASM] Document baseline WASM features

[SingleAccretion]

Contributes to #122309.

This is just the very initial version, I expect we will add more features as the codegen work progresses (e. g. BigInt integration).

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

[SingleAccretion]

@dotnet/jit-contrib

[kg]

We need EH and bigint and SIMD for sure, though I don't know which revision of EH we need/want.

[SingleAccretion]

We need EH and bigint and SIMD for sure, though I don't know which revision of EH we need/want.

Do we really want SIMD as baseline? It means WasmEnableSIMD == false won't work / will force-disable AOT.

which revision of EH

We could start with the exnref version and see if anyone complains. Currently it's not yet widespread but hopefully in 2 years that'll change.

[kg]

We need EH and bigint and SIMD for sure, though I don't know which revision of EH we need/want.

Do we really want SIMD as baseline? It means WasmEnableSIMD == false won't work / will force-disable AOT.

Performance without SIMD for mono-wasm was borderline unacceptable, to the point that we had to implement a subset of it in the interpreter, not just in AOT. So I suspect we will need it. I'm not against making it optional if it's important for some particular customer or use case - I was under the impression support for it was widespread by now. From looking at https://webassembly.org/features/ it seems like everything except Owi has had it for a while.

[SingleAccretion]

I don't have concrete data besides that one "MVP-only" engine user, so have added both EH and SIMD. The baseline can be adjusted down in the future if needed, though it tends to be harder to do that post-factum.

[AndyAyersMS]

Think we should just leave this open until we get most of codegen implemented.