[Bug] Vec\range handles ints which don't roundtrip through floats wrong

[lexidor]

Describe the bug
When the arguments $start and $end to Vec\range(...) are ints exceeding the max safe integer for a 64-bit double, this function may return a vec (list-like array) of length one because of floating point imprecision.

To Reproduce

var_dump(Vec\range(1 << 60, (1 << 60) + 8));

array(1) {
  [0]=>
  int(1152921504606846976)
}

Expected behavior

array(9) {
  [0]=>
  int(1152921504606846976)
  [1]=>
  int(1152921504606846977)
  [2]=>
  int(1152921504606846978)
  [3]=>
  int(1152921504606846979)
  [4]=>
  int(1152921504606846980)
  [5]=>
  int(1152921504606846981)
  [6]=>
  int(1152921504606846982)
  [7]=>
  int(1152921504606846983)
  [8]=>
  int(1152921504606846984)
}

Environment (please complete the following information):

• OS: 3v4l.org (Linux?)
• PHP version: 8.0 through 8.2.10
• Version: 2.7

Additional context
Casting an int to a float may truncate the number. So this if() considers some large integers (or hugely negative integers) to be equal when they are not.

    if ((float) $start === (float) $end) {
        return [$start];
    }