Check definitions of arguments before suggest move expr out of loop

[xizheyin]

Fixes #141880

The first commit is used to add tests and the second shows the changes. For the test cases I added, there are three cases, and only the third case passes compilation after applying this suggestion. The changed code implements this.

r? compiler

[fee1-dead]

If I search for the string "consider moving the expression out of the loop so it is only moved once" from the codebase, I haven't seen a single suggestion that is correct in the context. I think it might just be better to remove this.

cc @compiler-errors if you have opinions

[fee1-dead]

@rustbot author

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[xizheyin]

There does seem to be something werid about this suggestion. As I tried it, I realized that the triggers to make this suggestion correct seemed a bit harsh, but it was indeed the correct suggestion in this instance:
https://github.com/rust-lang/rust/blob/master/tests/ui/moves/recreating-value-in-loop-condition.rs
https://github.com/rust-lang/rust/blob/master/tests/ui/moves/recreating-value-in-loop-condition.stderr
What do you think about it?

@rustbot ready

[fee1-dead]

I still don't think this is the correct approach. The original suggestion is very limited in what kinds of contexts would trigger a suggestion. For example,

fn foo() {
    let vec = vec!["one", "two", "three"];
    while let Some(item) = vec.into_iter().next() {
        println!("{:?}", item);
    }
}

This above doesn't trigger the suggestion even though it should. We have many false negatives and false positives. This should be rewritten by someone experienced but I don't have enough time/effort at the moment.

r? compiler

[xizheyin]

I also agree that this usage scenario is very limited. Do we need to remove it now?

Maybe I can try to rewrite it, but I don't know if the difficulty is acceptable to me.