Simplify Slice File Hashing Logic

[InsertCreativityHere]

This PR simplifies the logic used to compute hashes of our Slice files.
No more extension trait, with multiple implementations; just one function.

The only change in logic is that previously, we would compute a hash for each file separately, then hash all those hashes together:

total_hash = hash(hash(file1), hash(file2), hash(file3), ...)  // old

Now, we just have a single hash engine, which we directly feed all the files into:

total_hash = hash(file1, file2, file3, ...)  // new

I can't think of any adverse effects to this simpler approach, compared to the previous nested-hashing approach.

[externl]

I forget where were landed on the discussion of only including src files. Was this previously being done elsewhere?

[InsertCreativityHere]

Yes, you're right I should of added a comment mentioning this.
Currently, we do the filtering in slicec-cs:

let src_files: Vec<&SliceFile> = compilation_state.files.iter().filter(|f| f.is_source).collect();
let hash = SliceFile::compute_sha256_hash(&src_files);

https://github.com/icerpc/icerpc-csharp/blob/9edecc0ccaddee829828f2da64ce150b692d340f/tools/slicec-cs/src/main.rs#L65-L67
My thinking was to move it in slicec, to keep all the hashing logic and what/how we do it in a single place.

---

I'm also not completely sold on one vs the other, but this is the current behavior we have, so I kept it.
I'll blog about it, and see if we can reach a conclusion.

[pepone]

Much simpler, nice!