MOS is obvious, for three-terminal devices: source is common to substrate, so you get a diode from S to D.
Four terminal MOSFETs have gate vs. "back gate" (substrate) ambiguity, but that is easily resolved (substrate acts like a JFET gate, since after all, it's a diode junction, not MOS). Such devices are very old and (as far as I know) exclusive to RF, so may have symmetrical D/S construction.
A lot of the RF MOSFETs are designed for closely controlled impedances (think striplines) so probably are highly symmetrical. Not my territory really, I rarely go anywhere even remotely near the black arts of RF.
Why would that encourage symmetry? As Vgs is what matters, there's probably priority to having lower ESR and ESL on that pin.
Most RF parts are designed for, well, whatever they fit into -- after all, a 50V 10A transistor isn't going to drive a 50 ohm transmission line very optimally! Indeed, terminal impedances tend to be extremely low, a consequence of it being easier to make a transistor "wider" (more cells / surface area / perimeter in parallel) than "taller" (higher voltage, lighter doping). This is reflected on package and circuit design, where wide flat terminals are used to keep the impedances low. A matching network solves the rest, give or take how much bandwidth the application needs (narrow band: LC or microstrip; wideband: transformers or esoteric structures).
MMICs, and transistors made for similar purposes, tend to be easily matched (or prematched, or simply made for) 50 ohm (or near that) loads, though.
So, not that you've needed RF, nor probably ever will, but now you know...
Tim
Tim