So I wanted to add a note here on my progress on my machine (not a Robotdigg, but similar), hopefully to help others.
I am currently running some 4% SnPb flux-cored wire. That seems to have solved a lot of the wetting. I'll work on switching back to SAC305 in the future once I obtain some with a bit more aggressive flux.
I've come to the realization that the main difficulty (beyond the wetting) is reliably controlling the amount of solder being applied to the joint. Because there is no feedback as to how much solder each joint consumes, the only option is to feed a certain amount of wire per joint. If for whatever reason a given board takes more/less solder than the last one to solder, you end up with more solder than you need leading to balls and bridges, or less solder than you need leading to dry solder joints. Sometimes on the same board.
This is particularly true if for some reason you have a single joint which doesn't wet. Then the solder ends up left on the tip for the next joint, which then ends up with a big ball, or bridging between a couple of pins.
My current thinking is to switch from the cleaning the tip once at the start of the board which I currently have set up to cleaning multiple times during the process, possibly a lot of times during the process. My theory is that I should be able to make the process more deterministic that way and eliminate early failures from affecting later ones.