1).  Sure.  I’ve made Programs that simulate a POG architecture.  For all 4 octaves at once, plus input unison, you won’t get the EHX slow attack or filter-type attacks.  And in your use case, I’d use two PolyPhony instances in parallel.

You ‘can’ use a single QuadraVox instance to generate all 4 octaves plus input, but the processing is less forgiving.  Especially with you using mic’ed, complex input like wind instruments.

In Series, two PolyPhony instances can generate some very intense octaving.   But you are ‘processing the processing’ 2X, so the polyphonic accuracy starts to suffer.  More organ-esque.

2).  Any of the (currently 62) algorithms, two of them at a time.  Loaded in Series, or in Parallel.  With the option to Insert another two inputs, mono or in stereo, anywhere in that signal chain.

If you haven’t yet, check out the online manual.  It’s really complete, and well done:

https://cdn.eventideaudio.com/manuals/h90/1.3.0/index.html