Add dataset "splitter" that adds "with limit n"
We have users who want to run an analysis and limit the number of files by picking the number of jobs -- but if there is a recovery configured, it will tend to launch a recovery for the whole rest of the dataset...
So we need to make it convenient for them to limit to N files; it seems to be too much trouble for them to declare it themselves...
We could then make this the default for sample analysis jobs(?).