Fix generic bounds on buffer mapping

[kvark]

Smaller (and incomplete!) alternative to #119 while it's still discussed.
One missing piece here is alignment. cc @Coder-256

Also, importing wgpu-core as just "core" wasn't the best idea 😅 : it collides with the actual core.

[grovesNL]

bors r+

[Coder-256]

I think this solves part of the issue, but why not just return a &[u8]? I'm not convinced that would be the best long-term solution, but it would partially fix the problem of alignment (since [u8] has an alignment of 1). Users can initialize their data from manual offsets with from_ne_bytes, which was recently stabilized for floats (so it should be available on 1.40).

[bors]

Build succeeded

• continuous-integration/travis-ci/push

[kvark]

@Coder-256 that would address the alignment flaw, yes, but it seems super inconvenient for the user, which I am currently being while dogfooding the library. It might be the way to proceed, I'm just not in a rush to call that while the very question of "Should we have mapping in WebGPU?" is being actively discussed by the working group.

If you want to help by making a PR that removes the generic stuff from mapping altogether (as well as drops zerocopy dependency), while providing some way or example for the users to operate (!), I'd be happy to merge it.

[Coder-256]

@kvark I agree 100%, it's very inconvenient, but sadly I think it might be the only option. See my latest comment on #119. I'll start work on a PR. To be honest, I'm not 100% sure how it will turn out but I'll give it a shot and see how it affects the examples.