One further thing, I don't think CP is used to reduce multipath for FPV, it's to reduce polarisation mismatch fading. I assume you're using linear polarisation on one antenna and CP on the other? That being the case, you already have a 3dB loss. However you are much less likely to see deep fades due to polarisation mismatch.
AFAIK most people use CP on both ends for multipath rejection. However, I'm sure some do what you suggest when doing FPV aerobatics.
Hmm, I'd have thought polarisation mismatch more than multipath.
If both antennas were CP, multipath would still be a problem as reflected CP will be in the opposite polarisation sense, nulling the direct path.
But surely the CP rx antenna and the reflection are cross polarised and so the reflection is attenuated?
On reflection(!) you are absolutely right. The reflected reverse sense is strongly attenuated due to the polarisation mismatch, it does not destructively interfere. CP at both ends makes sense for mitigating multipath in this case.
One thing to consider is that the further the drone is away, the lower the elevation angle for a given static height. My understanding is that there's a 500' AGL flying limit (at least here in the UK), so realistically having helicals or patches pointing up and down with (say) a 45 degree beamwidths isn't going to be great when the a/c is a little way away: you want greater gain at the lower angles, given inverse square law. Plus there's the complication that off-axis, the polarisation won't be very circular, and in some cases will even be in the opposite sense. Directly overhead, the distance is at its minimum, you don't need much gain overhead.
A Lindenblad is a good idea, but fabricating one for 5GHz will be mechanically challenging.
As a matter of interest are all FPV video links analogue? I can see why of you're trying to minimise latency of you're controlling the a/c, but I thought drones were self flying, or is this RC? If latency isn't a problem, a FEC digital stream would typically provide a better range and be less prone to multipath fading, but it would innevitably introduce latency as part of the temporal diversity scheme introduced to mitigate against burst interference and fading.
Sorry, I have no direct experience of drones and RC. (FWIW, I fly real aeroplanes ;-) )