As I understand the CPS operation, the ICU counts 14 holes after the sensor detects the void in the flywheel ring, and fires when #1 is lined up, then counts 28 holes and fires again, presumably in time for #3. When the void passes again, it counts 14 holes, fires when #4 is aligned, then 28 holes later, fires again, presumably in time for #2.

But what happens if the sensor picks up a stray void (can't see the holes, which the ICU interprets as a void) from a damaged or mis-aligned flywheel ring? My guess is that it starts counting to 14 again, and fires at a time when the distributor rotor is not lined up with anything. Consequently, it will fire #1 and #4 consistently on the detection of the real void, but fire out of time for #3 and #2 on a false void.

The fact that you're getting fire on #1 and #4 suggests that the CPS is correctly reading the machined void, and correctly counting the 14 holes to the next firing sequence. Which suggests that the CPS is doing its job.

You might be able to detect out of sequence firings by putting a timing light on the center coil wire, and looking for random, out of time blinks.