use headers to determine padded size for outer malloc routines

_From [derek.br...@gmail.com](https://code.google.com/u/117968039472581148324/) on December 10, 2010 17:58:27_

PR 595800

PR 476805 adds interception of malloc routines in every app library, and
for outer routines on Windows we only have _msize() which returns the
requested size. To get the padded size today I'm guessing by aligning:
this case covers reading the headers to get the actual size. My code does
read the Rtl headers, that just needs to be hooked up to the outer layer:
figure out the right base.

_Original issue: http://code.google.com/p/drmemory/issues/detail?id=198_