Partitioning sky and fluxstd tables by input_catalog_id or version

The sky and fluxstd tables have more than billion rows. For such large tables, a single table is likely to have performance and maintenance issues. 

I'd like to implement partitioning to split sky and fluxstd table by either input_catalog_id or version. The former is recommended as integer is easier to handle, while the current pointing optimization code queries calibration objects by versions (as far as I understand). 

Even after the partitioning, Q3C indexes will work.