Actually, both ideas there work...I keep forgetting that Quadnic module, and I ought not to. It's definitely a different character than the analog-source Doepfer, comes in considerably cheaper (by $200!) and saves 6 hp of space. Downside, though, is that you lose individual modulation and sync patchpoints. The other similar solution that seems to be viable gets more pricey, though: a pair of Klavis Twin Waves. $498 for two, but they fit in 16 hp and have way more features than either the Quadnic or the Doepfer combined. It's almost Braids-ish, given the algorithm feature set. If the price weren't a barrier, honestly, I'd go in that direction. As for the Pico DSP, yeah, that could work as well, but if you're going to use it for that insert trick make sure it can do a decent job on a mono output. Of course, that module also need not be a effect processor; you could do some twisted things with waveshaping in that Chronoblob feedback loop, so that every delay iteration keeps getting more distorted and bent out of shape. A ring modulator would be neat as well, but you'd need a carrier source...even so, consider a delay in which the iterations get more and more clangorous. Lots of possibilities exist...
As for the MAX situation...there might be a solution in MAX that can emulate the additional Monome modules. MAX is so open-ended that if you're deep into it, there'll almost certainly be a way around any hardware limit. Or, explore some of the Expert Sleepers stuff to get the MAX patterns over USB (or even ADAT optical) to their hardware, piping the MAX patterns through Silent Way. At the very least, ES's stuff saves space and is equally open-ended as Monome's modules.