Despite my earlier comment (about the proprietary hardware) I don't think there really is a purely technical reason. Both the proprietary nature of Android and the Raspberry Pi may present some challenges to third parties, but I don't think that alone is sufficient explanation. I could be wrong about that as I only have a cursory idea of the details of this problem, but I do believe they are not insurmountable.1
I had it in my mind that there have been some functional normal (as opposed to IoT) Android builds over the years for various models, though I've never sat down to try one. More than seven years ago, when there was still just the model 1 A/B, the Foundation actually announced Android was coming but this fizzled out (unfortunately the links into the Foundation site from that Q&A are dead now because they did not maintain stable urls).
If none of the various attempts have produced an "Open Source build with a healthy user and developer community", it's essentially because there was not sufficient interest to provide the skilled developer hours, something for which there is a high demand especially in the volunteer sector.
Keep in mind here that the goal is perhaps not a very interesting one. "Normal user" Android (as opposed to IoT), pretty much by definition requires a display, ideally a touch display. By the time you are through with that, the power supply, an SD card, some kind of an enclosure, etc., you are looking at a device that costs $200+ US -- and it isn't even portable. You can buy cheap Android tablets with comparable (or better) specs, with a promise in writing that they will work out of the box, that are licensed by Google, and in a form factor that a Pi based device will never have.
While there are lots of feasible use cases for a stationary, light duty touchsceen based device that could be appropriately and cost effectively fulfilled by a Pi based device, these don't require Android and most developers of such would probably not want to tie themselves to that considering what else is available for them in this context.
So, realistically, you have something which demands a lot of work from people who have to decide what, if any, public projects they are going to put their free time into -- and for this one "the public" would not really promise much in the way of a "normal user" community. Which means it would require people who want to do it knowing it will mostly be for themselves, a realm where the competition is overwhelming: If I am going to work coding most of the week then sit down on the weekend to do more purely for fun/personal interest, getting Android working on the Raspberry Pi is just not going to be be the first (or second, or third...) choice of most of the people who actually have the skills to pull it off.
- An issue related to this is that you can't effectively install the Play Store except on devices licensed by Google, and that will likely never include the Raspberry Pi. This is not actually any kind of blockade to running Android, but it may be part of the reason why most people would not want to use it.
