ewertb

@ewertb, This whole answer isn't necessarily to address your set up, but to rather address panel positioning in general. Your " less than ideal" orientation and numerous different aspects is not a disadvantage, and no you don't need numerous smaller MPPT's if you do things right.
I'll explain:
The maximum voltage threshold that an MPPT can deal with, will dictate the number of panels you can have in a string.
Every panel in this string should be of identical spec and pointing in the same direction at the same tilt.
Now the current capability of your MPPT will dictate the number of strings you can deal with.
So you can parallel your similar strings to match this current capability.
Now, I want you to be aware of something, a solar panel with very little light still basically outputs full voltage. It may not be capable of supplying current at low light but its operational curve is such that it is very close to full voltage. This in turn means that it wont drag down the voltage of a panel with good light that can output a decent current. Which means a shaded panel can be paralleled with an un-shaded panel with impunity.
The common misunderstanding I am trying to address, is that it is often thought that all strings to a single MPPT have to all be at the same tilt and direction as each other.
As long as each string contains the same number of identical panels, and within each different string the panels are the same tilt and direction, then the strings can be paralleled.
Between the strings themselves, they can be at a different tilt and direction to each other on the same MPPT, without an operational detriment to speak of, in fact this has a considerable upside.
The tilt and direction of a string determines when in the day a string has its peak output. If all the strings pointed in the same direction they would peak together, and that peak current versus your MPPT's capability would determine the number of panels you could have. Your MPPT will have to be sized to deal with a big peak of current for a short period of the day.
This is inefficient, because the other 92% of the solar day, that MPPT is idling. The cost of that MPPT is a production overhead.
Paralleling strings with different directions means you can match waxing and waning strings together and spread their peaks. This softens the noon peak.
This in turns means you could either use a smaller (cheaper) MPPT, or alternatively increase the number of panels and get more useable units of power throughout the day. 
The MPPT still isn't overstretched at noon, but at least it is working at a reasonable pace throughout the day.
I mix and match strings like this, and you'll be surprised how many more strings of panels you can add before you reach your MPPT's capability. Paralleling low tilt E and W strings, anecdotally, I estimate it is about twice the number of panels, of an ideal tilt N - facing array. My loads are such that I can use every watt I make in a day, in other words my MPPT's do not back off when then batteries are charged.
You might argue that this is just trading panel efficiency for MPPT efficiency.
So, I'd like to point out two further advantages that I have noticed:
1.Getting the power throughout a longer solar day also makes the power more useable than having schedule my loads to match a massive noon peak.
2. On an overcast day, when you really need the power, a panel will still produce say 10% of its output, largely regardless of it's tilt and direction. Having twice as many panels under those circumstances makes a welcome difference.