i have used both methods, im not keen on relays, i used one on the bias boards because it allows grounded base operation when the bias is turned off, the relay need not be used if a transistor switch is used,
the darlington sounds a good idea, it only needs a little sniff of rectified rf which should not mess up the input vswr,
a transistor switch is potentially more reliable, if its switching power to a regulated circuit theres no problems with vdrop or thermal effects like you see with none regulated circuits,
chasing stiff regulated biasing is maybe overkill, its nice to have if maximum linearity is your goal but i cannot hear the difference between that and simpler circuits so long as the bias never dips low enough to cause severe none linearity or switching distortion,
i have used and modified several different circuits from simple switched pass transistor fed from a regulator to the modified motorola design, i even tried unregulated but thermal tracked circuits, they all work ok but the mot design gives stable bias voltage under any supply voltage/drive levels and pretty stable idle current under a wide range of temparature,
it can be setup to current limit swinging back towards class c at a certain drive level which may give some degree of protection if the amp is overdriven,
it also completely avoids the horrible texas star and messenger like problems of having the amp biased too hard in order to maintain a reasonable level of bias under high drive conditions,
with active regulation the amplifier can run cooler while remaining in the linear region,
the ultimate bias circuit imho would do the above and automatically compensate for changes in collector voltage on the biased transistors