I believe PyEphem is completely scientific. He uses LibAstro and XEphem, so I believe that it is as accurate as they are, which is very accurate.
You are definitely golden for your centuries around 1900 A.D. In his Halley Comet example, Brandon Rhodes uses year 1066. See fixed-objects-precession-and-epochs .
I found your question because I was curious about the epochs and years in ancient history. If you look at my github problem on a similar question, you will see that I got the same results as some pro-astronomers 100,000 years ago and 100,000 years in the future.
Hope this answers your question :)
source share