Merge pull request #2072 from RosettaCommons/roccomoretti/rs_include_fix
Attempt to fix behavior of read_in_and_recursively_replace_includes()
Brian Weitzner observed that RosettaScripts was **SLOW** with very large XMLs. He identified read_in_and_recursively_replace_includes() as the holdup. The previous version was using a rather sub-optimal algorithm, with a bunch of wasted effort. This PR changes the algorithm used to a more efficient one. Instead of exponential-ish behavior with XML file size, we now have linear-ish behavior.