I've experimented with placing GMR angle sensors outside the housing of a single pole-pair BLDC inrunner motor to try and measure the commutation angle directly from the rotor magnet. With a lookup table, I was able to recover enough of a signal to get the motor to spin, but very inefficiently. The housing and field lines produced in the stator distorted and interfered with the signal significantly.
Could you, in theory, get a magnetic field sensor to work outside a stepper motor as a crude rotor position sensor? Probably. Will it work well? Probably not. It depends strongly on the motor and sensor you're using, and making sense of the signal you (may or may not) get would be difficult, and would likely be totally different if you tried to repeat the setup.
It's pretty easy to test for yourself, I recommend just getting a linear hall effect or TMR sensor, throw the output voltage(s) on an oscilloscope, and see if you can make sense of the result. It's a fun thing to try.