Input Driven Observations for a GLM-HMM unable to have >1 observation dimension

[Toby-Shaw]

As I was going through @zashwood 's tutorial notebook ("2b-Input-Driven-Observations-(GLM-HMM)"), I noticed that a section is incorrect. At the start of section 2, it mentions adding an additional observation dimension if that was relevant for the given task, however this is not possible in the current ssm library, with several hardcoded assertions preventing multidimensional observational data. Sadly, my lab's data sort of requires two observation outputs to fully capture behavior, and there isn't a good way to combine them or finagle it. Is there any way to have multiple observation dimensions for an input-driven GLM-HMM?

Relevant Error at lines 689, 695, 736, and 759 in observations.py: "InputDrivenObservations written for D = 1!"