tpch performance testing improvements

Noting the task items so they don't get lost

1 - looks like our default tpch test configuration is aimed for large datasets, but in our benchmarks we almost never use that large sets.
2 - due to this data distribution we can only support q1, q6, and q12 for real time tracker, the rest fall into task tracker
3 - we can support more queries in real-time executor if we can make customer table as reference table (it has 1.5 M rows in SF=10)

Therefore we could do following
1 - keep the current configuration as it is an use it for large scale factor testing (30+)
2 - create a new distribution configuration that has reference table for customers.

In addition we did not document our postgres configuration when we run tpch testing. we shoulds
1 - document key postgres config values : shared_buffers, work_mem, max_parallel_workers_per_gather ...
2 - document data loading times in addition to query times
3 - tabulate query running times instead of providing raw data that makes it harder to follow