Itertools::max is misdetected as Ord::max, resulting in expected 1 argument, found 0 error

[TonalidadeHidrica]

Steps to reproduce

Create a new bin crate, and replace src/main.rs with the following code.

src/main.rs

fn main() {
    let _ = vec![vec![1, 2, 3]]
        .iter()
        .flat_map(|x| x.iter().flat_map(|x| Some(x)))
        .max()
        .unwrap();
}

When edition = "2021", nothing is wrong.
When edition = "2018", rust-analyzer points out that there is an error on the fifth line (.max()): [rust-analyzer mismatched-arg-count] [E] expected 1 argument, found 0.
Hovering on the max shows core::cmp::Ord, not std::iter::Iterator.

In both case, cargo check shows no errors or warnings.

Probably there is an issue in resolving traits in prelude.

Version info

rustup show

active toolchain
----------------

stable-x86_64-apple-darwin (default)
rustc 1.56.0 (09c42c458 2021-10-18)

:CocInfo

## versions

vim version: NVIM v0.6.0-dev+nightly-756-g8c74c895b
node version: v14.18.1
coc.nvim version: 0.0.80-03c9add7cd
coc.nvim directory: /Users/username/.vim/dein/.cache/init.vim/.dein
term: tmux
platform: darwin

[TonalidadeHidrica]

I forgot to mention, but I guess I was using 2021-10-25 then. Now I updated to [coc.nvim] rust-analyzer 04f03a360 2021-11-01 stable. Then I found similar error, even in Edition 2021. I was looking for how to reproduce it, then suddenly everything went well -- even for the one I explained initially!! I have no idea why it happened, but since I cannot reproduce it anymore, I'll close this issue. Please reopen if some could reproduce it.

[lnicola]

No, this is a real bug, I can reproduce it in 1.57.0-beta.2, but only in the 2018 edition.

[johnmave126]

I can reproduce a similar issue in 1.56.1 with 2021-11-01 rust-analyzer in the 2021 edition by shadowing try_into with a custom trait. See an example at rustcc

Submitted as a separated issue #10677

[flodiebold]

@johnmave126 would you mind reporting that as a separate issue, including the code example and what the problem is? It's not obvious it's actually the same problem, misresolved methods can have many underlying causes.