Any software project inevitably has many limiting factors. Some that spring to mind are available time and expertise, competing priorities, usability, and hardware capabilities.
It's hard for anyone outside of Sequential to provide much accurate insight into any of these with the possible exception of usability. The front panel of the Prophet X isn't going to change, and it's intentionally designed as largely knob-per-function. So there are limits to how usable any new additions can be, and it's likely they won't stray terribly far from the initial concept. I guess you could think of granular as a different type of looping mode, so let's give it a passing grade there.
As for available time and talent, Sequential has shown a pattern of returning to instruments to refresh them periodically during their production life, so I would hope that holds here. We've seen one major update already, so I'm hopeful that we will see more to come when the work fits into their overall schedule. Right now, I expect their focus is very much on whatever new instrument is next in the pipeline.
Priorities are just about impossible to guess. I don't doubt they listen to feedback, so this thread almost certainly weighs in, but they also have direct requests from customers and the never-ending backlog of minor bugs that could be squashed. I trust they'll put their time and effort where it seems likely to yield the best results, and know that being closer to the project than any of us gives them a broader perspective than any single customer.
On the hardware front it's very much a guessing game. I would guess, however, that there's not a ton of computing power that isn't already spoken for. The instrument is doing real-time transposition of a maximum of 32 looping stereo samples today. It feels quite optimistic to hope that it would suddenly do eight times as much sample playback work to get a really basic form of granular (as originally conceived, granular synthesis was proposed as consisting of hundreds of granules - so the Quantum's eight granule approach is already pretty sparse.) Perhaps lowering transposition quality could make something like this possible? I couldn't say for certain, but I wouldn't hold my breath.