I have some (seemingly common, but specifically unanswered) concerns about providing a neutral wire to smart switches that are in a single multi-gang box (4-6 switches on different circuits). I am wondering if it is acceptable to use a single neutral line run directly from the electrical panel to serve all the switches, even though they are on separate circuits with their own neutrals at the fixtures.
See diagram of what I am thinking below. 
This question relates to this and this question (to which the answers were "not ok"), but the difference being that I am not "stealing" a neutral from another nearby outlet or switch, but running an extra neutral directly from the panel.
To better answer @ThreePhaseEel , the setup is a bit out of the ordinary, and the current (inherited) physical setup is like this:
Best Answer
Keep 'em separated
Since you have two subpanels feeding this setup (one for each "side"/half), you'll need to make sure that the switches for each side have a neutral return back to the corresponding subpanel, and only the corresponding subpanel. In other words, the two left-side switches have a neutral wire going back to the left-side subpanel, and the two right-side switches have a neutral wire going back to the right-side subpanel.
Cross-coupling the two smart-switch-neutrals in your setup violates 300.3(B)/310.10(H) and creates a neutral current loop that causes inductive heating of metal parts (like your raceways and boxes) as well as magnetic field EMI due to the loop area of the current (old-school picture tubes were susceptible to this in particular).
Neither of these neutrals needs to join with the light fixture neutrals on its side until the subpanel neutral bar, though. The smart-switches are loads of their own (to power the "brains"), and running an extra neutral homerun in the same cable, conduit, or raceway is OK by code, as per 300.3(B) (otherwise, /2/2 NM wouldn't be Code legal, for that matter):
Load isn't a concern here, but Code is
The primary concern for multiplexed neutrals is having multiple hots on the same phase return via the same neutral, which causes the currents on that neutral to be the sum of the currents on the various hots. In this situation, though, that isn't a concern as smart switches don't draw much power at all.
However, technically speaking, unless you punch the circuits down as such to have currents cancel (i.e. a MWBC -- such things do exist in the three phase wye world, whether they are two-phase/three-wire or three-phase/four-wire), your proposal violates 200.4(A):