mmap/unix.rs: Add with_minimum_alignment to MmapRegionBuilder

[jaredwhitedev]

Add an option to specify a minimal alignment for the mmap'd region. This is useful to ensure VM memory backings get huge EPT entries.

Summary of the PR

Please summarize here why the changes in this PR are needed.
To get huge EPT entries callers must be able to align their memory to larger page sizes, such as 2 MiB or 1 GiB.
Using huge EPT entries increases the memory performance by decreasing the TLB pressure.

Expose a with_minimum_alignment option to MmapRegionBuilder to allow callers to do so.

To support this, we first allocate size+alignment_size of virtual address space, then use MAP_FIXED to place the actual mapping at an aligned address within the reserved VA space. We then free the excess

Requirements

Before submitting your PR, please make sure you addressed the following
requirements:

• All commits in this PR have Signed-Off-By trailers (with
  git commit -s), and the commit message has max 60 characters for the
  summary and max 75 characters for each description line.
• All added/changed functionality has a corresponding unit/integration
  test.
• All added/changed public-facing functionality has entries in the "Upcoming
  Release" section of CHANGELOG.md (if no such section exists, please create one).
• Any newly added unsafe code is properly documented.

[jaredwhitedev]

Seems that we can't allocate with more than a 512 MiB alignment when using Miri. Would you rather we:
a) Try best effort for Miri (Let the allocation failure happen whenever caller uses more than 512 MiB alignment). Current behavior.
b) Don't support Miri when using with_alignment (explicit error)
c) Ignore it when using Miri (default 8 byte alignment)

[liuw]

Non-conformant values are rejected in the build function, but I think it will be better to reject them here. This function can return a Result<Self>. In the error case you return InvalidAlignment.

This is a cleaner approach. Users of the alignment value don't need to validate it again.

[jaredwhitedev]

Updated

[andreeaflorescu]

@jaredwhitedev are you still interested in getting this PR merged? If yes, could you please re-open it in https://github.com/rust-vmm/rust-vmm/. We moved vm-memory there.