you may be able to get away with removing temperature altogether and just using density. I would also suggest upping the dissipation for density so you don't have an enormous plume of steam.
You can also dial a lot of the look in in comp, so don't be afraid to render "denser/bigger" than you need and play with the opacity after the fact.
if you change the hair type to splines, can you then cache the whole lot out as an alembic then render the outer tentacles with a S&T mat?
EDIT: have you tried caching the hair too?
This package looks useful and would be a nice simple way to get "troublesome" geo into Unity. https://assetstore.unity.com/packages/tools/modeling/mega-cache-26522
Your best shot is to report the bugs on the RS forums. Make sure to provide a log, steps to reproduce the crash, and if possible a cut down version of the scene.
I'm aware this isn't necessarily useful for _you_, but here's my 2p: the shortest path script works on the inside of a piece of temp geometry, in this case a spline/sweep which is in inside a volume object. Pretty handy for making things like veins, or tree trunks/branches.
In the timeline you can show/hide keyframes based on takes. Bear in mind you won't be able to actually edit them unless you drag and drop the specific object you want to edit into the take you want to edit it in.
EDIT - here's the menu
You won't be able to export visibility from C4D. Use scale set to absolute, uniform, -1 instead. That should do the trick. You'll probably also want to get rid of any falloff on the field so that stuff within it is immediately "removed".
I suppose the question here is: do you need to simulate this? the geometry and setup looks simple enough that you could just rig this up with either a pose morph or some deformers.
Yes. That's the "powerslider" btw, in case you're looking for it in the manual. To select keys, you click and drag the area directly underneath the frame numbers.
EDIT: too slow!
What sort of images are you making? Product/architecture/vfx etc
What pixel size/ final image size?
What optimisation are you already undertaking? Retopo/LOD/rigs etc
What hardware?
What renderer?
If the blend shapes are Apple ARKit then I have a couple of Xpresso rigs which can be used to retarget and mix takes. Let me know if you're interested. In future I am looking to develop these further so that specific face regions can be blended across takes, which should hopefully make things like adding blinks, smiles, twitches etc a lot easier.