Remove the redundant MDExceptionCodeLinux enum and use the values from minidump-common instead

[gabrielesvelto]

No description provided.

[Jake-Shadle]

So interesting, there seems to be a difference between 1.64 and 1.65 on Windows, I could build and test successfully on Windows with this PR on 1.64.0 but then switched to 1.65.0 and got the same crash as CI did, so at least it's fully reproducible.

[Jake-Shadle]

So it looks like the dump_current_process test crashes inside RtlCaptureContext, I've actually had issues with that function in the past, I think I've found a workaround for now which is to build and run test in release, but I'll add an issue to look into this for later, but considering that code path only gets hit in the not-recommended-use-at-your-own-risk dump_local_context I wouldn't consider it super high priority, though still worrying that (seemingly) a rust update changed...something.

[gabrielesvelto]

That function does magic stuff with the stack so it's possible LLVM is hitting some edge-case where it generates incompatible code. Anyway WDYT about that clippy warning? I suppose we could Box<> the (large) errors but it would change the external API so I didn't want to go there just to appease it (even though it would have other benefits).

[Jake-Shadle]

I just wanted to unblock this change for now so I silenced it, but I think before the next release of this crate it would be good to actually amp up the clippy lints and actually address all of them or at least have better reasoning for why certain ones get silenced since overall it makes things better.