I wanted to do the same Matt and as you clarified, I get it that it's not possible. Rather I am taking a new approach now. The whole purpose was to insert records parallel or fastest way possible. So here is my plan:
Create a partition function with date as partition
Create a partition scheme with all on primary
Create the target table as partitioned over the partitioned scheme
Now run the processing procedures in parallel to populate data into temp tables for different years
And at end switch table to target partitioned table.
Do you see any loophole in this ?