Change states to unknowns

[TorkelE]

Changes mentions of "states" to "unknowns".
Changes get_states (and similar functions) to get_unknowns (and similar functions.

I have tried to go through the full repo, and ensure that states such as in "steady state", or "the state of the system" are fine. Realistically, once all of this is over, I will probably have to read through the docs from start to finish to check that it reads consistently and nice. However, this should mostly be fine.

We cannot check whether this passes CI until MTK v9 is released. Also, CI will fail until we have figured SciML/ModelingToolkit.jl#2459 out.

We may/may not want to merge this into #757, but figured it would be easier to keep things somewhat separated for now.

[TorkelE]

I have fixed the all_unitless deprecation issue here as well (turning this to the PR that allows you to actually create ReactionSystems using the new MTK version).

Now, we do all(u == 1.0 for u in ModelingToolkit.get_unit([unknowns; ps; iv])) to check whether all of [unknowns; ps; iv] are unitless.

[TorkelE]

I think this is ready now, tried to update the MTK thing. Tests fail, but at least Catalyst compiles. Once: #749 we should be able to look at this now.

I try to change "state" to "unknown" where it make sense. At some point when we are done with all of this, we'll have to go through from start to finish and ensure that we are consistent with what is a state, species, unknown, variable, and whatever.

[ChrisRackauckas]

Tests fail though, it needs the complete calls.

[isaacsas]

Should be ok to merge after addressing these minor comments. (I didn't check the changes within tests.)

[TorkelE]

Tests fail though, it needs the complete calls.

We plan to handle this in a separate PR (on way), this one is needed to actually compile Catalyst though, hence we are sorting it out first.

[isaacsas]

Then just merge when this is done, irrespective of the tests failing.