I definitely think this title should be in EA as soon as possible. Bug-hunting, figuring out what modders actually want, etc. will take a year or so.
Yes, a year. Not a couple of months to basically just beta-test. I hope they're wise enough to do it slow and go for the long game, sales-wise.
Just based on the features we now know are in, I'm a little worried that they don't know what the modding community's expectations are; too much stuff sounds like it's hard-coded, there's been no coherent discussion of callbacks, no real explanation of the workflows for many things that need to be clear at release (like, putting in a new custom sword that will show up in the shops should be a specific tutorial, same with armor workflows, imo) etc., etc.
I could care less about game-wipes, although I honestly expect the engine won't have all of the issues we saw in Warband, where (for example) data objects couldn't be merely re-ordered in the bytecode without causing a problem, because they used ints to store object references, instead of being sensible and using a string (even back when the engine was made, that was Premature Optimization at its finest). Things like this need to be caught early and fixed, if they want the game to be a hit with dozens of big mods supporting sales into the far future (which is what we'd all like to see).
I know I'll buy it when it comes out, but honestly, based on what we've seen thus far, I'm afraid it'll be Warband, but with fewer ways to mod it deeply and a much more complicated asset workflow, which sounds like a Fallout 4 type of game (that I wouldn't want to mod; pretty much all of the big mods here are effectively TCs or major conversions of Vanilla).