I doubt Keil chose the caret (^) because of its meaning as an operator in C. They also use it in combination with an sfr name which does not equal its address but its value under normal conditions. I suspect they chose the caret (^) without considering its meaning in the C language. Just like they chose to use the assignment operator (=) for setting the address. And Keil compatibility is the only reason I can think of for Infineon to try this with SDCC and document it in this note.

With SDCC I can think of several other ways to calculate the address of an sbit:

sbit at 0xD0 | 7 CY;
sbit at 0xD0 + 7 CY;
sbit at &PSW + 7 CY;

I very much doubt Keil calculates the address based on the used operator. What will it do with this?
sbit DUMMY = 0xD0^8;

Or this?
sbit CY = 0xD0+7;