Potential improvement to reversed_empty_ranges

[pzmarzly]

While it is stated that this lint cannot catch loops over dynamically defined ranges, it could be improved to catch the following common pattern:

let x = 5u32;
for i in x..0 { ... }

Note that it is a range over unsigned values, so it makes no sense for the right side to be zero (as the range will always be empty then). Similar check could be done for signed integers by iXXX::MIN const.

[mxxo]

I don't want to hijack this issue, but clippy also misses adapters on invalid ranges like:

// skips loop 
for i in (10..0).rev() {
         println!("{}", i);
}

[Zalathar]

I ran into a similar issue with a backwards loop that modifies a vector:

// Wrong!
for i in (vec.len() - 1)..=0 {
    // Code that structurally modifies vec,
    // so we need backwards traversal
    // and can’t use a slice iterator.
}

The correct range is (0..vec.len()).rev(), which also conveniently avoids unsigned overflow when vec is empty.

[giraffate]

reverse_range_loop is now included in reversed_empty_ranges, so I changed this title.