Given the form factor they pretty much have to be but patch antennas can be circularly or linearly polarized. Consider the GPS antenna in your SOLO. It's circular. Also see
2.4 GHz 8 dBi LH Circular Polarized Patch Antenna - 12in N-Female Connector - HG2409PCL-NF. The fact that the case is slightly rectangular suggests that they are linearly polarized but we'd have to take the cover off to be sure.
Also, wifi base stations don't fly through the air, so advice on how to maximize throughput on an office or home wifi network is not always applicable.
The only advice being given here is as how to best close the link. This does result in higher throughput as fewer packets need to be resent (in either direction) but the goal here isn't higher throughput - it is greater link margin. That's an engineering term that may not be familiar but the meaning should be obvious. With higher link margin you have a better chance of not dropping the link if you fly behind a tree or over a hill and the link will stay closed at longer range.
Because the alphas are directional antennas they cover a narrow angle of the ski than omnis (but they punch further). Therefore you DO NOT want them parallel.
. Assuming the video to be correct (i.e. that one is for the uplink and one for the downlink) you definitely do want them parallel for the simple reason you want the transmitter at each end pointed at the receiver at the other end for both links. To return to your flashlight analogy: you don't have two flashlights at each end. You have one at each end and a detector at each end. You want the flashlights pointed at the detectors and the detectors pointed at the flashlights. The antenna axes should be parallel.
Now if you have evidence that this is a diversity system please pass that along as I can't find any. The cards are made by some Latvian company and the only info I can get on them is the type of connector and the remark in the video that the two ports are for Tx and Rx. In a diversity system the radio optimally combines the inputs from the two ports irrespective of the pointing. In that case if you have two 10 dbi antennas both pointed right at the drone you would get effectively 13 dbi gain out of the array but you could take less than that if you wanted broader coverage in solid angle at the expense of broader coverage in range.
You want them canted to increase the area of coverage to compensate for their directionality.
As just explained that reduces the area of coverage because you are throwing away margin.
The antennas on the Solo are NOT circular polarized.
I can't find the post that suggested they were but no, they are linear and that does bring up another point. As I said in #8 the e-fields from the Alphas must be aligned with the antennas on the drone. These are in the legs which are not quite parallel. Thus the Alphas should be rotated as in the 4th picture but with the lines between the SMA and the L's in 'Alpha' at 5 o'clock and 6 o'clock assuming the E direction is from SMA to L. If it is perpendicular to that then the L's should be at 2 o'clock and 10 o'clock.
You ought to be able to figure out which orientation is correct by linking up to the drone with it on a non conductive table or bench facing you and far enough away that RSSI is not against the rail. Rotate one of the antennas about the SMA connector. If you don't see any change in RSSI as you do this rotate the other. If you don't see any change there either then the antennas are circular and it doesn't matter which way you orient them (as long as they are both pointed at the SOLO.) If you do see a drop then slowly adjust the antenna which shows it to minimum RSSI. You are now cross polarized. Rotate the antenna 90 ° from the cross polarized position and set the other one to be symmetrical.
