I used Altiverb around the time it came out, and was really impressed with the implementation and bundled IRs.
I don't know what the percentages of studio use between convolution vs algorithmic would be, but yea, algorithmic reverbs are favored for the reasons you stated. At the end of the day, for many engineers, it's less about the realism of spaces, and more about how it sits in the mix. Also, as with synthesizers, it hasn't been about the clarinet (or whatever) it was trying to emulate 75 years ago.
Also, with convolution, as long as they have good general audio specs, there isn't a specific "personality" of the convolver to sell, over another one-- it's all down to the IRs. Yes, better convolvers offer features like ability to change the pitch and envelope characteristics. IMHO, a good parametric EQ is kind of a bare minimum.
Logic & Mainstage have Space Designer (along with the bundled Impulse Response Utility); AudioThing's Fog Convolver is affordable (currently on sale), WaveArts's free Colvology XT is good, there are others.
Given the capabilities of long-IR convolution, I'm curious why Boss, Line6, Source Audio, Zoom, or some other fx companies didn't want to touch this tech. How well did the EPSi sell? Are Poly FX & Tasty Chips "boutique" enough that their selling-like-hotcakes stats aren't indicative of broader demand? I think if people heard what is possible, they'd lose their minds. I've lost mine.