vim's source code is really hard to test. Refactoring is needed to make code testable. That's true for every project that doesn't have tests yet. Another thing that makes it really hard to contribute to vim is that it supports operating systems that don't even exist anymore. I really don't feel like spinning a VM to make sure my changes don't break it on MS-DOS. I would rather send a PR that removes MS-DOS support than support that.
vim7 was a great improvement, and when I show people the time travel capabilities it blows their minds. But that was a long time ago. With the number of people to whom vim is very important, I'm surprised there isn't more results coming out of it.
Though maybe it is mostly that the core doesn't need to change and innovation can happen in the plugins? That is certainly true to some extent, but the async stuff was way overdue.
Regarding cost. Most developers need to be paid. I read [here](https://groups.yahoo.com/neo/groups/vimannounce/conversation...) that Google provided (some) funding to Bram. With neovim initially there were lots of new things (clean-up, async, terminal), I don't see a problem if there should be a bit of slowdown now. It's a team. I think it's robust. I prefer Neovim and occasionally use MacVim too, thus grateful to Bram also. No need for frustration imho (but I'm glad there is neovim:)
No, the core contributors have not been outright in their criticisms, but IMO the fork itself shows a lack of respect for Bran's work and a lack of willingness to work with him to make Vim better. And now tarruda has stepped back from NeoVim, perhaps recognizing himself the cost that Bram bears.
By simply contributing to the tests - one of the outright critiques of Vim itself - they could have brought a lot more confidence to accepting such patches to Vim, but they chose to take their ball and play elsewhere instead.
What frustrates me the most is that instead of two branches of Vim that are moderately better (with the future of one of the branches being left behind), we could have had one which was much better.