Unexpected behaviour when using FM with fixed phase

[cardinal-bin]

Describe the bug
When having one channel FM another, both with a fixed phase, I would expect that over a number of cycles determined by their relative frequencies the pattern should be the same. However, small irregularities occur, most noticeable with square waves.

To Reproduce
Set channel 1 & 2 phase to 180 (both starting at the same time on a reset), both waves to square. Set channel 1 frequency = /4, morph to -25 (i.e. a 1 beat pulse every 4 beats). Set channel 2 frequency = x1, FM from channel 1, linear, 25.

Expected behavior
Channel 2 frequency should jump to 4x, sending 4 pulses within channel 1's pulse, then drop back to 1x and send another pulse on the beat.

In fact the behaviour is inconsistent & a little unpredictable. Sometimes the FM seems to reach channel 2 "in time" for the frequency to be 4 x during channel 1's pulse, but sometimes it seems to be slightly delayed, and the result sounds like it could be out of phase.

Media
This video shows the above setup on channels 7 & 8 instead of 1 & 2. Both channels are opening VCAs with separate square waves so the effect can be heard - the faster channel controls a higher pitched wave.

Juniper.zip

Versioning (please complete the following information):

• Hardware revision 08
• Firmware version 0.5.3
• SD Card brand / model - verbatim 32gb (as supplied)

Additional context
Possibly related to #74

[ost-ing]

Will investigate this for 0.7.0