Probably much better than an 'actual' switch as there are no mechanical parts to wear out. LJ12A3-4-Z/BX is used extensively in industrial process automation. It would be interesting to see some actual figures on the accuracy of sensing for both inductive and mechanical switches, especially in a swarfy environment. I am discounting Hall switches at the moment. For some reason I do not regard them as the best solution for positional sensing on a machine tool. (Although I would like to hear arguments for and against the case).

I did some testing on my gantry homing, putting a line on the stepper shaft and it came back to exactly* the same position every time, no matter how far away it homed from**.

Exactly* is an observation of angle of the line, not a measurement with a dti or similar.
** Mach3 homing would compensate for this, by doing an approach and back-off.