If someone posts a part (best) or even just the breadboard svg above as an svg I can run it through the script and see what it says (and manually look at it for issues). That said, if it makes parts easier to make for you I’d say use it. I remember a comment from one of our other part makers a year or so ago that he had looked at the python code for this with a view to improving it but that it was to complex (and I assume undocumented) to make any progress. A fair number of the parts in core have been created with one version or another of this script so I don’t know that there will be more complaints than usual but I also don’t know how many of the parts have been manually cleaned up after such conversions either by el-j or the author of the part .
That does look pretty odd! It looks like they had to angle it (very probably as a retrofit long after the original board was made) to be able to fit the POE transformer somewhere where it would physically fit and not short something else while still being low enough that you could stack a shield on top of it. I’d guess looking odd took last place to working (a sentiment I can agree with 
).
I somewhat suspect (but without data to back that up) that we will find the dup pins are more an artifact of copy/paste in Inkscape (I know I have done so before and had to edit the resulting connector values to maintain sequence) than the eagle script, but I may be wrong. There is also code in the source (I don’t at this point know if it is active or not) that will do kicad and I think eagle file conversion to fritzing. Its more than possible this is work in progress that isn’t actually running yet, but it seems to be there in some form. That said test data would be interesting.
Peter