a couple of issues in mcsoda

When I used mcsoda to test XDCR code, I found a couple of issues, which might not be critical, but I think it would be nice to fix it.

1) the number of items generated is approximate. For example, when I use this to generate load

lib/perf_engines/mcsoda.py localhost:12001 vbuckets=4 doc-gen=0 doc-cache=0 ratio-creates=1 ratio-sets=1 ratio-deletes=0 min-value-size=1,3 max-items=100000 exit-after-creates=1 prefix=XDCR

I expect 100K new items created but actually only 99994 items loaded. Looks like it is a rounding issue in math code to compute # of items.

2) When loading data, other than normal keys with prefix "XDCR", I also saw keys like "BIN_AAAAXXAAAA", see screenshot. How does that happen and is it counted as normal items?