If it's for 4-20mA signals, 0.5mm (20g) conductor is more than big enough. There would have to be more than a couple kilometers of cable before cable resistance would be a factor for typical instruments with a lift off voltage of 12-13Vdc, leaving up to 5V (for a typical 24Vdc loop power supply) for copper wire IR drop after accounting for a max 5V drop across the receiver input.
As to why there's two different cable sizes, maybe the instruments get wired by a different crew than those running the home run cables.
I know of a couple plants that don't use sizes smaller than 18g (0.75mm) because that's the minimum size for 'robustness'.