So I looked electronic compasses and the like and your post. Here is what I think the deal it. Prior to the P4 (might be P3... but at least up until the P2) there was one IMU in the Phantom. I think with the installation of two IMU's the system is able to self diagnose. I think this is done by comparing readings from the two IMUs. If they are not the same, the software will prompt the user to recalibrate. But, as you mentioned, DJI never removed the old information from the manuals. They eventually removed it from the Phantom manual but not the Mavic manual.
So I would agree with Meta4 and your post, that it does not need to be done unless prompted.
Thanks sar104, thanks Meta4.
In terms of calibrating the compass, the number of IMUs (or even compasses) is not the issue.
The compasses (3-axis magnetometers to be precise) simply measure the 3-D magnetic field. The single piece of data that the FC needs from the compass(es) is the direction of magnetic north relative to the aircraft, derived from the horizontal component of the earth's field that is aligned magnetic north-south. However, due to its ferromagnetic components, the aircraft has its own magnetic field, so that the field measured by the magnetometers is actually a linear sum of the two. The FC needs to know the aircraft's magnetic field so that it can subtract that from the measured magnetic field, leaving just the earth's magnetic field. It learns the aircraft's magnetic field by the calibration process, since that field is the one that doesn't change when the aircraft is rotated, unlike the earth's field that obviously appears to spin as the aircraft turns. That is all that calibration is for.
The longstanding misconception that calibration is to correct for magnetic declination is obviously incorrect. Magnetic declination is the location-dependent difference between magnetic north (that the magnetometer measurements provide once the aircraft's field has been subtracted) and true north. A compass, no matter how you rotate it, can never determine declination because true north is, magnetically, an arbitrary direction, and neither the compass nor any of the other sensors can determine the direction of true north without prior knowledge of the local declination. The local declination is actually provided by a global model of the earth's magnetic field that is loaded into the firmware.
The addition of multiple compasses is to compare their values, and might be useful to determine if one or both is out of calibration, but it does not change the conditions that require calibration, which are a change in the magnetic state of the aircraft.
Just for comparison, the P4 series has two IMUs and two compasses, the Mavic Pro has one IMU and two compasses, and the Mavic Air has two IMUs and one compass. The same calibration advice applies to all of them.