How's in-process issuing the same library much different than using a separate process? What's complicated there in monitoring/debugging that's different that's not otherwise?

I think that debugging with one less process is a good thing. And monitoring with one less process boundary to traverse is easier as well.