Regarding the cab's:
Sampling is a ardous process not only for the time but when you are capturing several variables especially distortion, it's a very sensitive process. To sample cab's properly you would need more than 10 distortion Kernels and presently that's tough to do without completely messing things up. The high freq's mentioned are most likely do to the sampling process interpolating "garbage"/noise within the signal since the Kernel implementation is so high. But when this is tackled properly - well, just look at what Nebula has done so far.
The included libraries that come with free Nebula as well as the paid versions are very usable indeed. The 3rd party libraries take it to unbelievable sonic heights. Aquavox too, don't forget is free, and is a way of wetting your beak so to speak. If the new user can understand and sort of step away from traditional coded plugin use, then they can begin to see that Nebula in general is about subltey and cummaltive processing. This is to say, don't push individual instances too hard and as you are building a mix you will notice in the grand scheme of things how much richer, more 3D, and organic your sound will be. Consequently, this also prevents form over processing like EQ and compression.