yes
yes (plus the general updates of the -stable kernel, which is not to be underestimated).
Not necessarily, if someone can provide a rather targetted fix for 4.14.x, chances aren't too bad that it will be applied - this depends on the impact these changes have, relative to the improvement.
Possibly (I don't own any mvebu devices, so I don't know).
yes, either for OpenWrt or upstream with the help of kernel.org upstream (this is not mutually exclusive).
Every fix should first go to master, depending on the impact (future maintainability of the patch, regression potential for other devices) and the gains, it may be backported to the release branches and end up in a maintenance release.
Not yet in the prebuilt snapshots, but it is available if you build it from source and enable TESTING_KERNEL (I think Davidc502's community builds include a v5.4 based image).
That is very, very hard to measure, as there are slightly over two dozen core developers with direct commit access, but many more semi-regular contributors - and even more feed maintainers (packages from the additional package feeds), but among these only a subset have access to mvebu hardware - so any value between one and ~150+ people might apply, depending on the details. Fixes can be submitted by anyone, not necessarily regular contributors; support for new devices is often submitted by non-regulars, who care about the device they've just bought and want to get it work.
No (it was roughly scheduled for summer, but it needs to stabilize first - and I'm not going to make predictions before it has been branched off master).
Yes, no - you'll have to follow development and administration mailing lists for that and ideally attend the IRC channels as well, basically when it's ready to ship with kernel v5.4 alone.
https://openwrt.org/docs/guide-developer/quickstart-build-images
There is way more detailed documentation as well, but that's the gist of it.