One of the virtues of the original Phase 45 is that it used a network of components around each of the two FETs that would reduce the risk of distortion. And because setting up a feedback path in a 2-stage phaser is rather pointless, it had none. So, on two counts the Phase 45 was diferent from its bigger brother: only one notch and no emphasis.
Its certainly not impossible for MXR to have built in all the necessary switching to convert the pedal from a true P90 to a true P45, but I doubt they did so. I don't know; I just doubt, largely because the 2N5952 transistors traditionally used for each pedal seem to only be available in full thru-hole form, rather than tinier surface-mount (the Behringer P90 clone uses almost all surface-mount parts except for the 2N5952s).
It's also the case that the script and block issues have different sweep widths and feedback. The later issue has a bit more feedback and a slightly wider sweep. In short, where the P45 and earlier script model is a little better suited to faster speeds, the later issue P90 is a little better suited to slower sweeps.
So, the push button switch that selects between 45 and 90 might be able to switch the LED and select between 2 and 4 stages. What the same type of DPDT button switches (which is what those switches are; I have a parts drawer full of them.) can accomplish with respect to converting from script to block, I don't know. It could switch the sweep width OR feedback, and LED, but not all 3 at once without electronic assistance and more parts stuffed into pedal.
I have no schematic or inside knowledge, so I'm speculating, but my educated guess is that the pedal does a partial conversion from script to block, or from P45 to P90. That doesn't make it a bad pedal in any sense. But if one is expecting EVERYTHING about a P45 and a script P90 and a block P90, I'm just not seeingit as feasible in a package using that size and those switches.