Yeah that's a pretty good idea.
Thinking it through as I type here, I think it would just be a modifier to the pit probe, a line underneath that allows the alarm value to auto-track the setpoint by a fixed value or percent. Then you'd probably want different percents for the high and low values, because you don't want it to alarm if it is about to go into lid mode? Also what happens if you're using not-the-first-probe as the control probe (pit probe disabled)? You'd probably want that modifier to apply to the control probe and not probe0. Then what if you set up your alarm values and set a setpoint but THEN disabled or enabled probe0, the alarm limits would be set on the wrong probe. Well that means they should probably be stored as the actual alarm values instead of as an additional modifier to just the pit probe which means the current storage isn't going to be able to hold it. So does that mean we start storing all alarm values as positive values and try to use 3 of the 16 bits as a bitfield, 1 bit for enabled, 1 bit for fixed offset, 1 bit for percentage offset, then the remaining bits for the value? That seems a bit hacky, but adding new probe parameters requires a new EEPROM layout and upgrade procedure.
Oh boy. This is why so little gets done in the codebase, because I start to think about all the possibilities and it becomes so complex that I never get it done.