[Bug tree-optimization/121921] New: `b - (b - e)` for pointers is not optimized to just e

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121921

            Bug ID: 121921
           Summary: `b - (b - e)` for pointers is not optimized to just e
           Product: gcc
           Version: 16.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: enhancement
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
  Target Milestone: ---

Take:
```
char *a(char *b, char *e)
{
  return e - (e - b);
}
```

This gets optimized to `return b` on the RTL level but I suspect it should be
optimized to b on gimple too.

Noticed this in PR 117924.

There we fully have:
```
  _201 = _36 + _35;
...
  _9 = _201 - _36;
  _10 = (long unsigned int) _9;
  _11 = -_10;
  _12 = _201 + _11;
```
(which does get optimized but only during forwprop4). But I suspect if we
optimize this earlier we could do the dse of the memset earlier there and then
dce of the `operator new/delete` earlier too.