Here are my five cents:

If you want true stereo, may I suggest using a stereo reverb instead. The Erbeverb has only one mono input afaik. It has stereo out, but it's not the same and may be limiting you.

Speaking of reverbs, they pair naturally with delays. In fact, a full blown stereo delay is the only thing I would be missing in your setup, if I wanted to do ambient.

As suggested by the others, a true stereo filter or a pair of identical filters does make a lot of sense for ambient. You could do nice stereo modulation with it. Although I would prefer a creamy 24dB filter for ambient. I would definitely try out the filter in person if possible. I have made the experience that you only really know if a filter or LPG is for you, by playing with it first.

If you want to handle sequencing only via midi, a Poly 2 or FH-2 will save a lot of space.

Hermod really shines when using it's sequencing effects section and when connecting the modular to external gear. However Hermods pianoroll is a little tedious. I would not recommend Hermod for sequencing without some kind of external midi input. If you plan on using it as a sequencer, I would definitely recommend trying it out first, as it may not be what you expect it to be.

And my final piece of advice, as you have mentioned yourself: The cool thing about modular is that you can start very small and learn as your system grows. When I got into modular I wanted to go all eastcoast, only to realize that I love westcoast. I would recommend getting a minimal voice first and expanding organically, constantly refining as you go along. However I agree with the others about getting a big enough case, even if it's almost empty at first.