`spatialJoin` handle newer default columns

[vpipkt]

currently the rf.spatialJoin(rf2) operation will clean up duplicates in the spatial and temporal keys, but not in related columns: bounds, timestamp

Although not explictly joining on those columns we can safely drop duplicated columns on either side of the join to avoid name collisions on the returned dataframe. Note spark does not mind the name duplication in the schema but it becomes impractical to query on those columns.