Guarantee repr(C) enum discriminants are stable within a compilation

[joshlf]

[joshlf]

cc @djkoloski @jswrenn

[joshlf]

Friendly ping 🙂

[joshlf]

Friendly ping 🙂

[ehuss]

Is there some context for this change? Has it been discussed by the lang team?

[joshlf]

Has it been discussed by the lang team?

No, but I figured it'd be pretty uncontroversial. Happy to CC them here if you'd like.

[traviscross]

@joshlf: Before we lang nominate it, could you update the PR description with a summary of the change being proposed here and describe in some detail the motivations for this change. E.g., if you could go into more detail about the situation you have with the zerocopy library and what the possible solutions for that problem are (especially if there are any other ones than the one proposed here), that would be helpful.

If you have any ideas for how we would actually encode this as a test in our test suite, that would be good to detail also.