i tink many may found funny and handy some features in recent 4.6 kernel.
like native MacSec support(https://git.kernel.org/torvalds/c/c09440f7dcb304002dfced8c0fea289eb25f2da0) , 5 version of BATMAN(https://www.open-mesh.org/projects/batman-adv/wiki/BATMAN_V), KCM (https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=9531ab65f4ec066a6e6617a08a293c60397a161b) and some - may find intel KPM useful(but not as more deep/serious counterparts, never accepted into upstream, yet), support for several new interfaces(both wired, wireless, fiber), support for SuperSpeedPlus USB 3.1 (10Gbps mode) and some others.
Yeah, we know. This is why v7 will get a new kernel (but not 4.6 AFAIK)
Changing Kernel will break many many things, this is why it can only happen in v7
I requested MACSec a few months ago. It would be great to see this in RouterOS v7.
yeah, toolchain consistency - mean lot of things. but sometimes certain/many of crucial things can’t be easy backported or cannot at all and thus - sometimes its only option.
my point is: i hope 7 version get “update enough” kernel anyway, even its not 4.6 yet.
normis above - already stated that unlikely, sadly 
maybe they reconsider it, because there was some time/efforts before we’re gonna see ROS 7, perhaps. but its unlikelym again. there was tradeoffs and schedule, always and Sometimes - someone had to puff trigger and release Something, even if its “not perfect. yet” thing 
its fantastic to had MacSec, PortSec and other 802.1x-2010 ajaced things in mainline/upstream kernel, without messing with bogus/buggy/poorly-documented patches/code from several different sources/platforms
thats why i like 4.6 changelog, among other changes in.
They can always backport the patch to whatever kernel v7 ships with.
depends.
“IF” they want to.
“IF” they had manpower/time/money for.
“IF” they consider that important;
generally keeping alive old branches - remain kinda expensive thing, so thats why relatively mid/big-sized companies(also depending on it)put in charge of backporting/testin/maintaining LTS/legacy things in.
its doable, sure. but in some cases - simpler to move upstream. trade-offs/balance - differ in Each case/product/company.
The other advantage of using a more recent kernel, are all the updated drivers.
Maybe the j1900 platforms out there, which make nice routers, can take advantage of it!
Any ETA on when the first Alpha/Beta comes out? I’ll be willing to test it. I don’t yet use my j1900 in production (well, It’s my hobby, and the J1900 is supposed to replace a bunch of RB493G boards).