wit: Migrate from float32/float64 to f32/f64

[sunfishcode]

Per the discussion in WebAssembly/component-model#277, there seems to be consensus for migrating the names of the types float32 and float64 to f32 and f64.

A plan for f32/f64 is:

• Make wit-parser accept f32 and f64 as aliases for float32 and float64. Accept f32/f64 as aliases for float32/float64. #1356
• Wait until that's in a release and the major tools have updated.
• Update Wit files, documentation, pretty-printers, etc. to f32/f64.
• After some time, add a deprecation warning for float32/float64 pointing users to f32/f64
• After some time, remove float32/float64

[alexcrichton]

#1364 implemented WIT_REQUIRE_F32_F64 as an env-var-based control for this, off-by-default for now.

[alexcrichton]

One perhaps unexpected piece of fallout in bytecodealliance/wit-bindgen#810 is that f32 is being promoted to a keyword with this change which means it's no longer valid to use as a function argument name. Not major, but figured I'd note it here.

[sunfishcode]

Ok, the wit-parser changes to accept f32/f64 have been in for a few months now, and wit-bindgen and other major tools have been updated and released, so we can now start on the next step, update Wit files, documentation, pretty-printers, etc. to f32/f64.

[alexcrichton]

Sounds good to me 👍