Open
Description
In the embedding step, when the user selects a column we could detect NULL values (and empty strings) and offer a way to deal with that.
One way would be to ignore those rows, marking them as "deleted" which we currently support in the code but don't expose in the UI.
The challenge may be that one column has nulls where another column doesn't. So on a per-scope basis its ok to have deleted rows but for the whole dataset you wouldn't want to assume that