more sophisticated replacement allocator: lockless thread-local allocation

_From [bruen...@google.com](https://code.google.com/u/109494838902877177630/) on July 20, 2012 13:28:07_

*\* TODO [#C] more sophisticated replacement allocator

split from issue #794 listing some sample optimizations here

**\* TODO add lockless thread-local allocation

however, this may complicate issues wrt the app lock and DR locks: xref issue #900 and DRi#779

**\* TODO add coalescing

note that an app that allocs a ton of one size, frees them all, and then
allocs a ton of a larger size will hit OOM w/o coalescing

**\* TODO add splitting of larger free block for smaller request

_Original issue: http://code.google.com/p/drmemory/issues/detail?id=948_