I'd not get too excited about this, if I were you. Unless you're happy to wait 10-20 years?
I'd suggest that most of what you're talking about could be done within Midi 1, it would just be VERY fiddly and maybe more trouble than it's worth.
All the midi sounds we have now are dependant on samples, usually modified using 'synth' type processes. There are cases now where multiple sample sounds, variants of the same instrument, have been made available. For example, a trumpet as you suggest. You could build a sound bank of as many different samples as you care to each representing a particular aspect of the sound, as you describe, even down to the same instrument but being played by different players using different styles. Then you could create a midi performance that uses a Program Change midi command tp select the required one, on a note by note basis. Fiddly? You'd need to do this pretty much manually, I'm not sure there's be any way that midi 'recording' could distinguish between the different sounds to do it automatically, as it can do with volume/velocity/frequency.
Maybe things will develop to allow more options with the 'synth' type processes, but I'd suggest that these are not THAT much to do with 'midi', either Spec 1 or the new Spec 2. Yes, the latter will allow many more conrollers, and might allow for more sensitive variation, but I don't see this doing much to help your complaints - the massive use of extra/alternate samples could help here.
Like I suggested, using alternate samples, layering, maybe lots of SYSEX to do 'synth' type extras, you could achieve a lot, but would it be worth the trouble?
I suppose the question I'm asking is 'how do you get this sort of detail INTO midi?' If you cannot get the detail IN, in a practical way, then is it worth thinking about how to get it OUT again?