adding knot size following disk profile

[matroxel]

Update adds non-zero sizes to the knot component of the profile, fixed to 250pc. Also changes distribution of knot locations to follow the disk profile, rather than a gaussian. A comparison of the impact of the changes follows.

Top: Version in main; Bottom: pull request update
Left: bulge component included; Right: Bulge component turned off.
(note, random seed not fixed between panels in test, so knot locations change)

[matroxel]

I've updated this pull request to include the calculation of n_knots as well and fix a bug where the random seed for the knot distribution was user-defined and not fixed, so different people using sky catalogs would typically end up with different knot positions.

If n_knots exists as a native attribute, it will use that value - otherwise it will use the new n_knots calculation, so if we want to add a more realistic fixed n_knot model later we can just add it to the catalogs and the default random code will be overwritten. Also enables backward compatibility with cosmodc2.

I couldn't figure out how to get a config-level entry inside the object classes, so the physical knot size (constant for all galaxies) is still hard-coded. I'm happy to change that to a config-able attribute if someone points me to how to get a value from the config and into e.g. base_object.

This should be fully compatible with current cosmodc2 objects and some things (like the random bug fix) should be applied to them no matter what, but still might be worth waiting to review this until after the diffsky additions are more advanced.