Ironically, in my experience this is a management issue.
In my games I use task groups with four gravsurveyors and one support/jump ship. I prefer jumping into the system, splitting up the TG into individual ships, and setting the surveyors to automatically survey. The problem is they don't always pick the most efficient path around the system.
Sometimes a pair will work their way around one of the rings of survey locations in opposite directions. When they meet on the other side they'll move out to the next ring, but they pick points that aren't adjacent. For instance, one goes to 21, then moves counterclockwise, and another goes to 23, then moves clockwise, leaving 22 to be mopped up later, typically by a ship that's on the far side of the system. Or the first ship to finish its laps will try to snag the last available survey point (again, Murphy's Law makes it the furthest distance possible) despite the fact that there's another ship at an adjacent point that is five minutes away from finishing up and scooting over to grab that last one.
I can micromanage the fleet, either assigning the surveys manually or watching them like a neurotic helicopter parent so I can correct their mistakes as they occur. But there's so much else to manage, like research and construction and mining and ship allocations. Personally I just accept that they're going to do it their way, and count the costs in time and fuel as the price of not concerning myself.
It would be nice if the AI were sophisticated enough for the ships to coordinate and say to each other "It's cool. I'll get this one. Don't hoof it all the way across the star system just for this straggler," or "Woops, we missed one. I'll knock that one out next so we don't have to come back this way later," but that's probably too much to ask. The AI would need to know my plans for where the TG will move to once the survey is over and factor that into some form of the Travelling Salesman Problem.
This kind of inefficiency is most easily avoided by having just one survey ship. Or embracing the micromanagement, but that's troublesome with multiple TGs in multiple far-flung systems. If you're planning on multi-year missions with minimal tanker support, a one-surveyor-per-system policy should reduce the fuel burden. If you want it done fast, you want more survey sensors in the system, and putting them on one hull ups the per-ship tonnage, which has its own costs.
