Fix df output on Windows

[lhecker]

This properly implements FsUsage using NT APIs. The choice
of GUID volume IDs is intentional as this matches PowerShell.

To assist with this, the PR introduces a new nt module,
which contains various helpers to interface with the NT API.
I expect their number to increase in the future, as the NT API
is rather well documented nowadays and expected to be stable.
They permit various implementations to be technically perfect,
such as the use of FILE_STAT_LX_INFORMATION.

Before:

Filesystem              Type  Size  Used Avail Use% Mounted on
\Device\HarddiskVolume3 NTFS  1.4T  749G     0 100%
\Device\HarddiskVolume4       730M  676M     0 100%
\Device\HarddiskVolume1       500M  168M     0 100%

After:

Filesystem                                        Type   Size  Used Avail Use% Mounted on
\\?\Volume{e7964102-1eaf-468b-a4df-dc8b143304b9}\ NTFS   1.4T  749G  625G  55% C:\
\\?\Volume{ac0578dc-5e8c-4f51-86e7-91bee5184973}\ NTFS   730M  676M   55M  93%
\\?\Volume{cc1ca74e-803a-4308-872c-d38f1157b9b2}\ FAT32  500M  168M  332M  34%

[DHowett]

If I’m honest, I have a minor preference for the short device node path rather than the volume GUID path; however, the volume guid is immediately accessible in an object browser, so… eh

[sylvestre]

it will need new tests

[sylvestre]

this is a lot of hardcoded values here
isn't that available in windows sys ?

[lhecker]

I didn't realize we exported NT symbols via windows_sys::Wdk. I based lots of this on my own C++ snippets. I'll make the necessary changes.

[github-actions]

GNU testsuite comparison:

Skipping an intermittent issue tests/cut/bounded-memory (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/tail/tail-n0f (passes in this run but fails in the 'main' branch)
Congrats! The gnu test tests/printf/printf-surprise is now passing!
Congrats! The gnu test tests/seq/seq-epipe is now passing!
Note: The gnu test tests/misc/write-errors was skipped on 'main' but is now failing.

[oech3]

Does this break MSYS2 compatibility if people prefer native df binary?

[lhecker]

I'm not sure what you - could you explain in more detail?
My understanding was that MSYS2 is its own thing, in particular because it uses POSIX paths. In this case, however, they also treat the Filesystem column fundamentally differently.

[lhecker]

BTW I apologize, but I'll need some more time to get back to this PR. I'm currently ironing out some other kinks at Microsoft. (1-2 weeks.)