Mike,
You are correct that powder does not "rust". However it does decompose to a red-brown material liberating acid which can corrode a metal can.
You are not correct that rust can be removed with a magnet. FeO2 is non-magnetic, and any particles that do stick to a magnet are due to metallic iron in the particles, not the red rust.
The IMR-4198 that went bad on me had a fine red brown dust in it and had the "acid" smell, not the normal ether smell. The metal cans were not corroded inside or out, so the red dust was not rust. I think the decomposition had just started & was not too far along.
When I tried it in fire forming some 30BR, the loads did not perform anything like the same loads with "good" 4198, really variable. I also didn't want to risk the fine dust corroding my Culver powder measure, so I trashed the 2 lbs of powder.
In 45 yrs of reloading, I've not seen this before.
Regards, Ron