Fix large rho0 vector size for BPHY23

We had fixed the large etac vector problem. However, in recent data processing: https://its.cern.ch/jira/browse/ATLASDPD-2126 we can still see some failed "Looping" jobs due to large rho0 candidate size. We added a new parameter to control it and tested against a few failed jobs, and found the jobs ran OK with the fix.