We can probably rule out a good handful of attempts that can't possibly finish within the search heap max time, just by tallying the minimum time necessary to finish the run. That is, we could sum up:
access times of required locations
the min access times of locations with required items (in case of duplicate locations)
traversal times to the spots with those locations.
Ideally, "required locations/items" is preparatory knowledge, but on the other hand, there may be additional requirements as locations are marked skipped during a route.
Alternatively, if no locations are skipped until the minimization phase, then we could just calculate the sum of all todo locations (taking the minimum of any canonical locations), as we won't even start that phase until we collect everything. It's unclear whether we should do a BFS to analyze the minimum traversal as well, or if we should just skip that.
We can probably rule out a good handful of attempts that can't possibly finish within the search heap max time, just by tallying the minimum time necessary to finish the run. That is, we could sum up:
Ideally, "required locations/items" is preparatory knowledge, but on the other hand, there may be additional requirements as locations are marked skipped during a route.
Alternatively, if no locations are skipped until the minimization phase, then we could just calculate the sum of all todo locations (taking the minimum of any canonical locations), as we won't even start that phase until we collect everything. It's unclear whether we should do a BFS to analyze the minimum traversal as well, or if we should just skip that.