[Bug tree-optimization/118483] Missed optimization due to cast being used more than once

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118483

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I am not sure we need the :S in the end.
I noticed the other patterns that do `a OP ~a` have an optional convert
assiocated with it: e.g.:
```
/* ~x | x -> -1 */
/* ~x ^ x -> -1 */
(for op (bit_ior bit_xor)
 (simplify
  (op:c (convert? @0) (convert? (maybe_bit_not @1)))
  (with { bool wascmp; }
   (if (types_match (TREE_TYPE (@0), TREE_TYPE (@1))
        && bitwise_inverted_equal_p (@0, @1, wascmp))
    (convert
     { wascmp
        ? constant_boolean_node (true, type)
        : build_all_ones_cst (TREE_TYPE (@0)); })))))

```
and 
```
/* Simplify ~X & X as zero.  */
(simplify
 (bit_and:c (convert? @0) (convert? (maybe_bit_not @1)))
 (with { bool wascmp; }
  (if (types_match (TREE_TYPE (@0), TREE_TYPE (@1))
       && bitwise_inverted_equal_p (@0, @1, wascmp))
   { wascmp ? constant_boolean_node (false, type) : build_zero_cst (type); })))
```

So we should be able to just do:
```
(for cmp (eq ne)
 (simplify
  (cmp:c (convert? @0) (convert? (maybe_bit_not @1)))
  (with { bool was_cmp; }
   (if (types_match (TREE_TYPE (@0), TREE_TYPE (@1))
        && bitwise_inverted_equal_p (@0, @1, wascmp))
    { constant_boolean_node (eqne == NE_EXPR, type); }))))
```

I will add/modify the pattern in 2 steps though.