xattr: move to platform pkg, use cython, use bytes, use fd

[ThomasWaldmann]

this code used to live in borg.xattr and used ctypes
(and was the only ctypes-using code in borg).

the low level code now was converted to cython and
the platform code moved to platform package.

got rid of the code that tried to find the libc.

[ThomasWaldmann]

Linux and macOS looks ok (see travis), will do other tests locally.

Please review.

[codecov-io]

Codecov Report

Merging #3918 into master will increase coverage by 0.23%.
The diff coverage is 82.17%.

@@            Coverage Diff             @@
##           master    #3918      +/-   ##
==========================================
+ Coverage   85.04%   85.28%   +0.23%     
==========================================
  Files          36       37       +1     
  Lines        9263     9185      -78     
  Branches     1545     1527      -18     
==========================================
- Hits         7878     7833      -45     
+ Misses        945      917      -28     
+ Partials      440      435       -5

Impacted Files	Coverage Δ
src/borg/archive.py	84% <100%> (+1.41%)	⬆️
src/borg/platform/__init__.py	90.47% <100%> (+1.58%)	⬆️
src/borg/xattr.py	67.21% <65.85%> (-15.96%)	⬇️
src/borg/platform/base.py	82.72% <71.42%> (-0.93%)	⬇️
src/borg/platform/xattr.py	88.88% <88.88%> (ø)
src/borg/archiver.py	85% <0%> (+0.39%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2178e73...34a51eb. Read the comment docs.

[ThomasWaldmann]

tried freebsd12:

• with fakeroot: lots of EISDIR fails all over the place (not xattr related)
• without fakeroot: "invalid argument" for listxattr('input/fusexattr')

[ThomasWaldmann]

about freebsd12 issues:

• the EISDIR issues seen with fakeroot are nothing new. freebsd10 testing in 1.1-maint just did not have fakeroot installed.
• the issues seen in fuse xattr tests seem to be a bug in freebsd12 fuse code, normal native fs works ok with xattrs. so, not a borg issue. also, the same issues happen with master code.

general issues:

• the currently hardcoded XATTR_FAKEROOT flag influences which tests are executed and how, likely this should get reviewed / improved / removed.

consider api change:

• getxattr returns None for empty value - check why it is not just b'' as one would expect.
• setxattr accepts a None value - check why it is not just b'' as one would expect.
• the latter 2 items somehow feel like if "None to represent empty" makes sense somehow, it is done at the wrong place (should be transformed one layer higher).
• listxattr returns list-of-str (requiring fs.decode() calls), could be list-of-X with X being "whatever getattr/setattr" on this platform wants
• getattr/setattr gets a str for the attr name, requiring a fs.encode() call, could be whatever needed (bytes)

cleanup:

• we already have borg.xattr.get_all, could also have borg.xattr.set_all
• get_all and set_all would deal with transforming native attr name datatype to str (for storage into borg Item) or from str (when restoring from a borg Item)