Fix unnecessary_lazy_eval suggestion applicability

[camsteffen]

changelog: Fix unnecessary_lazy_eval suggestion applicability when breaking type inference

Fixes #6240

[rust-highfive]

r? @llogiq

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

[llogiq]

I don't fully understand yet how the test is supposed to capture the difference in applicability. Wouldn't we need to have a //run-rustfix test for that?

[camsteffen]

Hmm I think you're right. The test doesn't verify that the lint is not machine applicable. But if I put the test case under //run-rustfix, the suggestion is applied in the .fixed file even though I have Applicability::MaybeIncorrect, and then the fixed file fails to compile. See the async_yields_async test which does the same except the suggestion does compile. Is this a shortcoming of the whole testing setup? What I did is consistent with tests assign_ops2 and bytecount.

[llogiq]

I don't have a good solution right now, but that shouldn't stop us. So... @bors r+

[bors]

📌 Commit a6611de has been approved by llogiq

[bors]

⌛ Testing commit a6611de with merge 4bbef42...

[bors]

☀️ Test successful - checks-action_dev_test, checks-action_remark_test, checks-action_test
Approved by: llogiq
Pushing 4bbef42 to master...