update to split dataset with replicates

[das046]

The idea is straightforward:

1. de-duplicate the dataset.
2. split and assign subset to each compound
3. map the subset assignment back to the original duplicated dataset.