I certainly wasn't talking about the phone version of Windows.

That's dead indeed. I mean regular desktop Windows 10 and Windows 7. To have Marvin run right there, so I can process annotations directly without even exporting them from Marvin, just like my friend does it with Google Play Books, would be a dream come true. It's likely never going to happen.
(I have a MacBook 13, but as I detest Apple software of all sorts, I only run Windows 10 on it, via Boot Camp. So I wouldn't really miss Marvin on Mac OS, because I don't use Mac OS.)
I'm also certainly not switching from Dropbox to iCloud as my main cloud provider. I also hear many developers complaining about iCloud, even in its latest iterations. You may potentially be in a minority that is happier with iCloud, but...

For example, the Swedish developer of a nice time-management app called ATracker was forced to switch syncing from iCloud to Dropbox, because iCloud wasn't being transparent enough for his programming purposes. The syncing was buggy all the time, whereas now with Dropbox syncing, everything is working fine.
I also have an Android phone and tablet, and iCloud would be useless there. Plus, there's a host of fantastic third-party apps that only really work with Dropbox – such as my cloud-based voice-recorder app, RecUp (formerly DropVox), or the aforementioned plain-text editor Permanote (formerly Nebulous Notes).
As to those $600 and $3000 upfront fees, I guess the developers would need to be transparent with their users about them. So, for example, it's September 2017 right now. The developer might ask their user-base: "Do you wish to continue using this app on platform A, B, C, and D in 2018? If yes, here's the cost."

A "fund-raising" webpage might be set up for each particular platform. I'm pretty sure that if, for example, all that it would take for us to make Marvin run on Android was to collect $600, we might be able to collect the $600 right here among MobileRead users only!

Perhaps I'm being overly optimistic, but if the $600 weren't collected on time – well, the app wouldn't run on that particular platform for the following year, then. This way, users would at least know what they're paying for.
I agree that paying for all those subscriptions for apps for everyday needs can easily get out-of-hand. Which is why I purchased a classic, one-time MS Office licence for both my desktop and laptop a couple of years ago, and which is why I'm not considering the purchase of a Permanote subscription, and would not consider purchasing a Marvin subscription unless Marvin starts at least to support the basic feature of highlights and annotations syncing (with further improvements in sight).