Add ValidationFlags::BINDINGS

[kvark]

Part of gfx-rs/wgpu#4409

A few things to think about here:

1. Is the scope of the flag chosen soundly? I was considering INTERFACE first, but it could be considered too wide of a scope.
2. Can this be released as a patch? I.e. is adding a bitflag a breaking change?

[teoxoy]

1. Is the scope of the flag chosen soundly? I was considering INTERFACE first, but it could be considered too wide of a scope.

What would INTERFACE include that the current BINDINGS doesn't?
(Is this question about naming or behavior?)

2. Can this be released as a patch? I.e. is adding a bitflag a breaking change?

I don't think so because a call to all() would include the newly added flag.

[kvark]

What would INTERFACE include that the current BINDINGS doesn't?
(Is this question about naming or behavior?)

kinda both, but BINDINGS is ok.

I don't think so because a call to all() would include the newly added flag.

Is this a problem though? the behavior of all would be the same, just the value is a bit different.

[teoxoy]

I don't think so because a call to all() would include the newly added flag.

Is this a problem though? the behavior of all would be the same, just the value is a bit different.

What I imagined could cause a problem is let's say as a user I call all() then proceed to "peel off" individual flags and then check if the bitflags is_empty().

This feels similar to rust's adding new enum variants (without non_exhaustive) which is also a breaking change.

[teoxoy]

LGTM

[kvark]

Ok, thanks for reviewing this @teoxoy !
I'm not blocked on this version being released as a patch, FYI, was just wondering if that's something to expect.