Allow funcref literals to have inexact types

[tlively]

Update the Literal constructors for funcrefs to take Type instead of
HeapType to allow them to be given inexact function references types
when the referenced function is an import. Use the new capability to
give references to imported functions inexact types in GUFA. Add a test
where this change fixes a misoptimization as well as tests where this
change simply changes the nature of the misoptimization. Future PRs will
fix these tests.

[tlively]

Will hold out on landing this until follow-ons are ready because --fuzz-exec is unhappy with the GUFA changes without corresponding interpreter changes.

[kripken]

I'll look into possibly landing this, if it helps fuzzing V8 in the medium term.

[tlively]

Meanwhile I've just uploaded the proposed fix upstream in the proposal repo: WebAssembly/custom-descriptors#72

[kripken]

I experimented here:

https://github.com/WebAssembly/binaryen/compare/main...kripken:binaryen:import.func.type?expand=1

That has most of the changes to make imported functions inexact, including literals and otherwise. A bunch remains, though - wasm-split, some C API questions, possibly more. Overall it is a bunch of work, and I'm not sure it's worth doing just for the short-term, if we and V8 are going to update to the final spec later anyhow.

That branch might be useful as a list of relevant places that the final spec update will need to modify (but differently).

[kripken]

I found a way to simplify the C/JS API issue, and also realized this is more urgent than I thought - we are probably missing quite a lot of fuzzer coverage atm, being out of sync with V8 - so I finished this up: #7993