compare Release /Od perf to /O2 and Debug natively and under DrMem

_From [bruen...@google.com](https://code.google.com/u/109494838902877177630/) on December 20, 2011 11:03:11_

We'd like to avoid tailcalls but still get good performance. We can disable inlining and FPO but not tailcalls in cl. We should measure Release /Od and see its perf. Full Release /O2 is 2.5x faster than Debug /Od natively. Release /Od should still be faster than Debug /Od b/c of no debug heap features from _DEBUG, and no DCHECKs (for chrome).

Xref issue #724 Xref https://code.google.com/p/chromium/issues/detail?id=108155

_Original issue: http://code.google.com/p/drmemory/issues/detail?id=725_