Convert scanner to C for great good (~˘▾˘)~

[414owen]

This should solve all past and future portability issues, because the C runtime is pretty ubiquitous.

The only generics we were using was in vectors, which I've preserved, with some macros.

C has no u32string, so I've used the vector macros to solve that too.

We lose namespaces ¯\(ツ)/¯.

The code is structurally the exact same.

We seem to have gained a bit of speed, compared to the C++ version (probably not statistically significant):

project	% parsed	time
effects	100.00%	0.0322s
postgrest	100.00%	0.0769s
ivory	100.00%	0.2119s
polysemy	98.75%	0.0807s
semantic	97.60%	1.0335s
haskell-language-server	99.79%	0.4648s

@wenkokke does this remove the need to patch tree-sitter for your wasm builds?
After this change, stdio.h is only required for -DDEBUG builds.

[tek]

sick! very curious to find out how I will deal with the C when adding features :grin:

[wenkokke]

@414owen realloc is not exported, so this would still require a patch.

[wenkokke]

Running the parser on haskell-language-server errors out with:

RuntimeError: Aborted(Assertion failed: undefined symbol `__assert_fail`. 
    perhaps a side module was not linked in?
    if this global was expected to arrive from a system library,
    try to build the MAIN_MODULE with EMCC_FORCE_STDLIBS=1 in the environment)

Which, I think, probably means that you have a failed assertion in there somewhere?

This aligns with my previous observations that the WASM tests would loop indefinitely on haskell-language-server.

[tek]

ugh why are the GA checks not running for PRs from forks

[wenkokke]

I'm working on patching the CI :)

[FoamScience]

I like the pure C too myself, been dealing with C++14 issues for a while now at more than one place!