What reason should we have to expect that there will always be the same number of different groups? splitapply() called that way is going to return one value per group that is found.
Also, findgroups() for numeric data is, in practice, going to create group numbers sorted by the numeric value associated with the group. If in one call the assorted values include (say) 0.1 0.11 0.2 and the next iteration the assorted values include (say) 0.12 0.11 0.2 then in the first iteration group 1 would be for 0.1 and group 2 would be for 0.11 -- but in the second iteration 0.11 would be group 1 instead of group 2. Therefore even if the values are related between iterations, the order might well change, which is likely to cause confusion.
