If you want to merge two datasets that could be parts of a whole with many transformations after that merge, there should be an option to replace the dataset at the intersection with the complete dataset so that the transformations that would be done on the merge can be applied on the full dataset.
Thank you for sharing your idea!
It sounds like you're referring to a scenario where two datasets are joined, and after this merge, you apply a sequence of transformations. You would then like the option to replace the two input datasets with a single complete dataset at the merge point, such that all downstream transformations apply to this unified dataset. Am I right?