MMap memory mapped file address allocation code cannot handle addresses non-aligned to memory mapped granularity size

As part of the fix to SERVER-10516, the code tries to find the next available free slot of virtual memory to allocate a memory mapped file. If the base address of the memory region or the allocation size returned by Virtual Query are not aligned to the memory mapped file granularity boundary, MongoDB will then try to map a file to an address not aligned with the correct granularity.

https://github.com/mongodb/mongo/commit/0205239ecb579e20b05cee3a0df394eb11ba14ce