[Bug tree-optimization/126042] -O2 -fno-trapping-math makes worse code than with trapping math for `<=>`

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126042

Drea Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|                            |106164

--- Comment #2 from Drea Pinski <pinskia at gcc dot gnu.org> ---
```
int f1(double i, double j)
{
  int t = i > j;
  int t1 = !__builtin_islessequal (i,j);
  return t & t1;
}
```

is just `i > j`

```
int f2(double i, double j)
{
  int t = i > j;
  int t1 = !__builtin_isgreaterequal (i,j);
  return t & t1;
}
```

is 0


```
int f3(double i, double j)
{
  int t = i > j;
  int t1 = !__builtin_islessequal (i,j);
  return t & t1;
}
```

is `i > j`.

But really combine_comparisons should be used. But that returns a tree. I am
going to change it to return a tree_code.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106164
[Bug 106164] (a > b) & (a >= b) does not get optimized until reassoc1