rebis-dev 1.77 and Clippy

[bakaq]

This runs cargo fmt (to make CI happy) and fixes the Clippy warnings (so that we can benefit from its advice without noise).

The diff for cargo fmt is quite big, so it may be better if you do this locally. Manually fixing the warnings wasn't trivial though.

Notice that the test that failed here is also failing in the current rebis-dev, so nothing regressed.

[Skgland]

Could you try cherry picking the commit from #2793 to fix CI?

[Skgland]

Mh, macos has a segmentation fault, nightly and 1.77 fail to compile and the rest have the same single test failure.

Edit: Its is wasm job that fails to compile not the normal nightly build.

[bakaq]

The single test failure is the one I mentioned in the OP. I've been able to get 1.77 to compile.

[adri326]

The tests should start passing again with #2759

[bakaq]

As rebis-dev was already cargo fmted (thanks @mthom!), I'm switching this to solve just Clippy and 1.77.

The tests have become flaky for some reason. I have run them a few times and sometimes they pass, but sometimes a test case fails for some reason, and it's not always the same test case that fails when this happens! I will investigate to see if this is a problem in rebis-dev or if my manual Clippy fixes broke something for some reason.

[bakaq]

Ok, the tests are indeed already flaky in rebis-dev. I think it may be because of UB in the unsafe code of the new Heap, because the panics and segfaults that are happening all seem to be happening in the heap. For example:

---- machine::lib_machine::tests::issue_2341 stdout ----
thread 'machine::lib_machine::tests::issue_2341' panicked at src/arena.rs:206:5:
value contains invalid bit pattern for field ArenaHeader.tag: InvalidBitPattern { invalid_bytes: 94 }
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

This looks like heap corruption caused by UB. It may be a good idea to write some tests for the new Heap and run them through Miri, because it will probably catch whatever UB is happening there if this is indeed the problem.