Test the laziness of our iterators

[Philippe-Cholet]

Related to #791
Add a test to:

• track iterators that are not lazy yet ;
• help track missing must_use attributes with very little discipline.

It's a long list of iterators and it has some edge cases. The error message help us understand how unlazy there are.

When let _ = ...; is not necessary, it indicates that a must_use attribute is missing.

CI will fail for some time.

[jswrenn]

Tremendous thanks for doing this work! There's no rush on this (as we have some time to go before we can merge this PR), but before merging, I'd like us to have distinct tests, rather than one big one. We can do so without too much trouble with a little macro magic:

mod laziness {
    struct Panicking;

    impl Iterator for Panicking {
        type Item = ();
        fn next(&mut self) -> Option<()> {
            panic!("iterator adaptor is not lazy")
        }
    }
    
    macro_rules! tests {
        ($(test $name:ident { $body:expr })*) => {
            $(
                #[deny(unused_must_use)]
                #[test]
                fn $name() {
                    let _ = $body;
                }
            )*
        }
    }
    
    use itertools::*;
    
    tests! {
        test interleave {
            Panicking.interleave(Panicking)
        }
        
        test interleave_shortest {
            Panicking.interleave_shortest(Panicking)
        }
        
        test intersperse {
            Panicking.intersperse(())
        }
        
        /* and so on */
    }
}

The advantage of this approach is that the tests are run in parallel, are reported individually, and can be run individually.

[Philippe-Cholet]

I surely had doubts about my big test. Thanks for your review.
I rewrote it.

Panicking: The ExactSizeIterator implementation is needed for circular_tuple_windows (the actual length does not matter). No DoubleEndedIterator implementation needed.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (b1be1e4) 94.40% compared to head (fee3f6d) 94.29%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #792      +/-   ##
==========================================
- Coverage   94.40%   94.29%   -0.11%     
==========================================
  Files          48       48              
  Lines        6665     6665              
==========================================
- Hits         6292     6285       -7     
- Misses        373      380       +7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Philippe-Cholet]

I'm not sure what to do about the remaining unlazy iterators yet.
I therefore added one commit to ignore the 3 failing laziness tests (then rebased).
I think we should merge this and un-ignore failing tests later.