Add un-vacuumed deleted rows #4
Description
Columnar data warehouses generally implement UPDATE by simply appending more rows and somehow flagging the "shadowed" rows as deleted. This benchmark would be more realistic if about 10% of the data in each table were deleted and replaced with similar data, so that there's some junk lying around that has to be skipped over by the query planner.