Fix several occurences of the improper_ctypes lint in tests where that was not the intent

[asquared31415]

Addresses most of #53858

Summary of changes:

allow(improper_ctypes) removed (it was not doing anything anymore)

• tests/ui/lint/lint-ctypes-fn.rs
• tests/ui/abi/variadic-ffi.rs
• tests/ui/abi/issues/issue-97463-broken-abi-leaked-uninit-data.rs
• tests/ui/abi/foreign/foreign-fn-with-byval.rs
• tests/debuginfo/type-names.rs
• tests/codegen/riscv-abi/riscv64-lp64-lp64f-lp64d-abi.rs (note: it's suspicious that there's an empty struct here that works, why is that the case?)

Lint removed by addressing the root problem

• tests/ui/issues/issue-3656.rs
• tests/ui/issues/issue-51907.rs
• tests/ui/repr/align-with-extern-c-fn.rs
• tests/ui/lint/clashing-extern-fn.rs
• tests/ui/lint/clashing-extern-fn-recursion.rs
• tests/ui/abi/issue-28676.rs
• tests/ui/abi/issues/issue-62350-sysv-neg-reg-counts.rs
• tests/ui/abi/extern/extern-pass-TwoU8s.rs
• tests/ui/abi/extern/extern-pass-TwoU16s.rs
• tests/ui/abi/extern/extern-pass-TwoU32s.rs
• tests/ui/abi/extern/extern-pass-TwoU64s.rs
• tests/ui/abi/extern/extern-return-TwoU8s.rs
• tests/ui/abi/extern/extern-return-TwoU16s.rs
• tests/ui/abi/extern/extern-return-TwoU32s.rs
• tests/ui/abi/extern/extern-return-TwoU64s.rs
• tests/ui/abi/abi-sysv64-register-usage.rs
• tests/ui/cfg/conditional-compile.rs
• tests/ui/abi/abi-sysv64-arg-passing.rs (note: one instance here is intentional in the test, can this be fixed?)
• tests/ui/lint/clashing-extern-fn.rs (note: two instances here are intentional)

Intentional or the lint firing/not firing is part of the test

• tests/ui/lint/warn-ctypes-inhibit.rs
• tests/ui/issues/issue-16441.rs
• tests/ui/lint/lint-ctypes.rs
• tests/ui/abi/extern/extern-pass-empty.rs (note: #![allow(improper_ctypes)] // FIXME: this test is inherently not FFI-safe.)

Unknown/Not addressed in this PR/Needs more investigation

• tests/ui/issues/issue-28600.rs (possibly intentional for ICE)
• tests/ui/issues/issue-26997.rs (possibly intentional for ICE)
• tests/ui/issues/issue-38763.rs (possibly intentional for ICE)
• tests/ui/issues/issue-5754.rs (possibly intentional for ICE)
• tests/ui/foreign/nil-decl-in-foreign.rs (maybe an LLVM bug test? It was fixed with no links to how.)
• tests/ui/mir/mir_cast_fn_ret.rs (I don't know enough about this test)
• tests/ui/mir/mir_codegen_calls.rs (I don't know enough about this test)
• tests/ui/abi/numbers-arithmetic/i128-ffi.rs (note: it's testing 128-bit ints in FFI, this is a known problem cc i128 / u128 are not compatible with C's definition. #54341 128-bit integers marked as having an unknown stable ABI #78473 )

[rustbot]

r? @petrochenkov

(rustbot has picked a reviewer for you, use r? to override)

[asquared31415]

@rustbot author whoops, meant to r? ghost in the first place, I want to do some more looking into some of the final list

[rustbot]

Failed to set assignee to ghost: invalid assignee

Note: Only org members, users with write permissions, or people who have commented on the PR may be assigned.

[petrochenkov]

You can keep me as a reviewer for this PR.
I'd expect majority of allow attributes in tests/ui to be removable, not only improper_ctypes, and that would generally improve quality of the tests, but keeping this PR focused on improper_ctypes is better because it will already become pretty large with this lint alone.

[bors]

☔ The latest upstream changes (presumably #108611) made this pull request unmergeable. Please resolve the merge conflicts.

[Dylan-DPC]

@asquared31415 any updates on this?