Good stuff! Small error: if I want to look at December 31st in a non-leap year in a sweep (e.g. 12/31/1990), the code computes the sweep stop as 365 + (sweepstep)/10, e.g. 365.1. The IRI 2012 interface rejects this.
I've removed the step/10 bit as a stopgap, but good to correct in later versions.