Add Rust impl of wasmtime_ssp_proc_exit

[kubkon]

Replaces C implementation of wasmtime_ssp_proc_exit with pure Rust implementation. This PR is the first one of a series addressing #120.

[sunfishcode]

Looks good, with one comment: Let's put the new implementations in a new file, rather than adding to host.rs. I know host.rs isn't very big right now, but its conceptual purpose is to provide types and constants and be the counterpart to wasm32.rs. Putting implementations in a separate file will help us keep the separate concerns of translating between guest/host data types and actual interface implementations separate.

As we move more things to Rust, we may eventually want to remove wasmtime_ssp.h altogether, and move what's left into host.rs -- which will be the types and the constants.

[kubkon]

Sounds good. I'm all for good organisation of code especially if, in the future, it means simpler and cleaner maintenance.

We could either put in a separate module at the crate level, say host_impls.rs, or we could make it a submodule of the host module. In the latter case, we could then use the "import-all" trick to hide the submodule from the rest of the crate and, as a result, get away with not having to change the internals of the actual syscalls in syscalls.rs. So, to sum up, we could have it as

// in host.rs
mod impls; // or some other, more descriptive name
pub use self::impls::*;
// ...

// in host/impls.rs
use super::*;

pub fn wasmtime_ssp_proc_exit(...)

with the code in syscalls.rs unchanged. What do you reckon? Which option do you prefer?

[sunfishcode]

I think I have a mild preference for having a new module at the crate level, rather than a submodule. It may even become an entirely new crate eventually.

[kubkon]

I've moved the implementation to host_impls.rs module at the crate level. Let me know if host_impls is a good name though!

[sunfishcode]

host_impls.rs works for now. Things will likely evolve over time, to be more modular, etc., but we can figure that out as we go.

[kubkon]

Hmm, analysing js-polyfill/build.sh (which I completely forgot about when submitting the PR), I think I might have been too hasty in removing the wasmtime_ssp_proc_exit from wasmtime_ssp.h. I believe that by doing so I've broken the polyfill as now __wasi_proc_exit will be missing from the exported functions table. Would you agree @sunfishcode? If so, let me know and I'll re-add the missing symbols to wasmtime_ssp.h, and posix.c. Sorry about this!