update future.{write,read} ABIs

[dicej]

This implements WebAssembly/component-model#524 insofar as it affects validation, componentization, etc.

• future.write now accepts its payload value as up to 4 flat parameters before spilling to linear memory.

• future.read takes no payload pointer when it has no payload type

• {stream,future}.close-{readable,writable} have been renamed to {stream,future}.drop-{readable,writable}

[dicej]

CI failures appear to be due to me using the latest Rust release; will downgrade to the MSRV locally and fix the issues.

[alexcrichton]

Looks good 👍

In the *.wast tests I think I only saw renamings, so could you add tests for futures for both reads/writes to ensure properties such as future.read has no pointer with no payload and future.write uses flat lowerings or a pointer if the flat lowering is too big? Additionally I think it'd be good to have a test that future.write doesn't require memory if the lowering fits

[dicej]

In the *.wast tests I think I only saw renamings, so could you add tests for futures for both reads/writes to ensure properties such as future.read has no pointer with no payload and future.write uses flat lowerings or a pointer if the flat lowering is too big? Additionally I think it'd be good to have a test that future.write doesn't require memory if the lowering fits

FWIW, I've tested those cases end-to-end in wasip3-prototyping, plus the fuzz tester is hitting those cases also. I agree we should have some handwritten WAST tests as well; I'll add those.

[alexcrichton]

Yeah all the code here looks correct so I don't expect any failures, mostly just handwritten tests for the well-known edge cases are good to have to catch any possible issues in future refactorings earlier