Next alpha release

I've finished implementing a patch for a bug found by a poster on comp.lang.c++ that could cause the RRef::isNull to erroneously think an object still existed, rather important :) Thanks to Ivan for pointing it out. The fix doesn't affect the public interface, but does have a small impact on performance (I have to repeat the performance tests) and memory footprint, but is more amenable to multi-threading if that ever happens.