The hard piece of your question to me is finding something that will both output MIDI or triggers/gates/modulation and also do audio input. Not impossible at all; in a way it’s helpful, because that limits your options a lot, which will help you make choices.
In terms of sending MIDI out from your iPad to modular, you can get a MIDI to CV module, of which there are many. I’ve been eyeing the Hexinverter Mutant Brain as a flexible option for percussion sequencing myself, but really there are many options here.
With this path, you also need to have an iPad interface that handles audio input and MIDI. I am not sure how many there are of these that are iPad compatible but certainly there must be some. Others may be able to chime in here.
The other way to do control would be to send your trigger and pitch information out as DC-coupled signals. This requires an ‘audio’ interface that is DC-coupled, of which there are not that many. The ES-8 from Expert Sleepers is the go-to option here, though there may be others. It is not cheap, but it offers 8 channels of signal out and 4 channels of signal in. These can be audio or CV. There are a number of videos showing Audulus controlling or being controlled by modular via an ES-8, and presumably this would work with other iPad software also. This would take care of your in-out problem without the need for two pieces of hardware.
In terms of a voice, your minimum ‘requirements’ are a VCO, a VCA or low-pass gate to control the loudness of the oscillator, and some sort of envelope to control the VCA/LPG and give shape to the sound. I put requirements in quotes because you can make sound without the VCA & envelope. But if you’re starting with making ‘notes’ that have duration and attack/decay shape, that is what you’d need.
The Mangrove is actually not a bad starting choice, because it has volume control built into it, saving you a module. That said, you could have a simple oscillator and VCA or LPG for less than the cost of a Mangrove.
If you go the ES-8 route (or other CV out of your iPad), you can generate your envelopes from your iPad and have your trigger from the iPad timed with it. However, I would actually recommend that you invest in a Make Noise Maths if you’re starting out. Maths will take care of your envelope needs and still do much else besides. It is a bit daunting looking, but actually in the end very simple, very flexible. It can do many things with its component parts, and is very ‘modular’ in spirit, if modular is the idea of combining simple building blocks in ways of your choosing to create complex behaviors. Maths really is all that. It’s also very well documented on YouTube and such, so you’ll have a lot of tutorials to get you going.
Semi-modulars are nice because they’ll give you all these things in one package: oscillator, VCA/LPG, envelopes, MIDI. I would give consideration to starting with one.
Good luck! Ask more questions 