Finally ordered the basic components for the auto-aiming wireless repeater bot, everything other than the servos and wheels basically :) Bought from Baltic Networks who had great prices and pre-sales support.
Basic platform is the Ubuiquiti RouterStation @ $63 (going to be running OpenWRT). Much more powerful than a usual robotics board (and much larger) but I'm definitely hoping that the GPIO pins are enough for the servo control.
Wireless cards were a pain in the ass, too many options to optimize :) In the end I went with the lower cost/lower power cards the MikroTik R52nM mini-PCI cards. I did want the DBii F20-Pro for the long-distance link (it's 500 mM vs 300 mW) but those are out of stock everywhere (most likely because they appear to approach the XR2 in power and tuning for long-distance but are half the price!). The R52nM are lower power but about the lowest power consumption I could find (allegedly 1.95W). Honestly, I probably could have chucked used/surplus cards in for the prototype, but I didn't have any and I'm not entirely sure about openWRT support for some of the "pulled out of laptop cards". These were $32 each, but even the cheapest aren't much less than that (the F20-Pro would have been between $50-$60 and the Ubuiquiti XR2 at least $90).
For the enclosure I went with the RF Elements StationBox with a built-in 2.4Ghz 14dbi panel antenna. Baltic Wireless suggested that it's going to be a bit tight in there, so I'm not entirely sure where the servo is going to go. Perhaps someday I'll upgrade to a higher dbi parabolic grid antenna and re-house everything, but they are heavy at 5.5lbs (good wind shear, though). The enclosure + antenna together were $23.50, so I think I did well price-wise there.
Other than that there was a PoE power supply (annoyingly I have one left on the boat that I can't get to.) $11. And a little omni-antenna to create the local side of the network (pigtail from MMCX to RP-SMA $5.75, 3dbi omni itself $10.75, but at least it's less than 10cm long).
Now the plan is to get all this working as a static repeater/bridge, figure out the weight and then robotize it (ie get it controlling a servo). The plan in my mind is to have it hanging from the top of a pole and have a kind of track around the pole into which the teeth of the servo mesh to drive it around the pole. The whole unit will move, rather than just the antenna and this doesn't account for tilt, but since most of the weight is taken on the top of the pole, I think this should be the least taxing on the servo/strongest. I'm a bit concerned as to what happens when the antenna needs to be aimed in such a way that it is against a strong wind, but deal with that when I come to it (ie the wind could blow back against the servo and burn it out?).