I don't know why I haven't been doing this for the last 10 years... In the past, I've used two different techniques - track the encore crowd noise as a separate track or leave it on the tail end of the last song prior to the encore. Recently, I started deleting it altogether and crossfading the encore into the rest of the set during the small amount of crowd noise I leave. It is surprisingly simple and results in a show that flows better - who needs the crowd noise?
Here is what I do (using Wavelab 6):
1. Start about 10 seconds prior to when the encore tracks start - usually there is some tuning/chatter, etc. I include that. I copy from that point to the end (to the clipboard).
2. About 10 seconds after the end of the main set, delete from there to the end of the show. (I have the encore music tracks in my clipboard.)
3. Highlight about the last 5 seconds of the main set, select Crossfade from the Process menu. I use the default values and Paste.
Results in about 10-15 seconds of crowd noise and the crossfades sounds seemless as it is down during the crowd, not any music. It takes about 30 seconds to do and I'm amazed I haven't done it until recently.