rms_region is poorly defined/documented

[lboogaard]

The rms_region keyword of findclumps is potentially confusing:

• it's docstring is unclear: it is a length (in fraction of the spatial cube size on each axis) that defines a central square region in the collapsed slice over which an rms is computed
• this rms is not actually used in findclumps (as sextractor does its own computation), but only provided in the catalog to have an estimate for the particular kernel/channel combination

In any case the docstring and should be clarfied. We could consider not even exposing this parameter to the user, or (perhaps better), making it more useful (for example, such that the user can define an aread that, for example, avoids the center).

[lboogaard]

Also: the default value of 0.25 (linear dimension) is quite small, as it leads to only 0.125 of the area being used.
The size of the area is in fact the center +/- rms_region, so it does scale as described.