Unfortunately it's not possible to edit the query after the dataset is created. It needs to be re-created. We've had requests to change it but it's never been at the top of anyone's list. I still expect that to be fixed.

Yes, I think you will have to do the pipeline import, though you could do a mix of the two if you wanted.

As for the Oracle data source, glad to hear you are taking a crack at it. Let us know if you need any help. Starting with MySqlDialect seems like a good way to go, though I think syntactically Oracle may be closer to the pg dialect.