While looking into where we use the psfShape slots in DM-35060, I noticed that ComputeExposureSummaryStatsConfig sets the starShape and psfShape fields to base_SdssShape. Given the discussion on RFC-857, we probably want to be using HsmShape pretty much everywhere, although that RFC is about characterize, and the input to ComputeExposureSummaryStatsTask comes from calibrate. That means we probably should default those to use the slot_shape and slot_psfShape fields instead.

It doesn't look like the test of this task checks the psfStats output, so we might have to use pipelines_check or ci_hsc to confirm that the change is ok.