MRG: skip precision test on macOS, newer numpy

[matthew-brett]

Numpy 1.12 blacklisted the macOS powl function, meaning that the use
of power functions for testing maxexp on float128 doesn't work on macOS.

[coveralls]

Coverage increased (+0.0003%) to 96.255% when pulling 4cd533b on matthew-brett:mac-precision-fix into 3d492a6 on nipy:master.

[effigies]

Merging upstream/master to make sure that the test failures are the expected ones.

[codecov-io]

Codecov Report

Merging #583 into master will decrease coverage by <.01%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master     #583      +/-   ##
==========================================
- Coverage   94.32%   94.32%   -0.01%     
==========================================
  Files         177      177              
  Lines       24695    24697       +2     
  Branches     2639     2640       +1     
==========================================
+ Hits        23294    23295       +1     
  Misses        925      925              
- Partials      476      477       +1

Impacted Files	Coverage Δ
nibabel/tests/test_floating.py	86.59% <66.66%> (-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 0d02553...ee7183e. Read the comment docs.

[coveralls]

Coverage increased (+0.0003%) to 96.255% when pulling d68339c on matthew-brett:mac-precision-fix into 0d02553 on nipy:master.

[effigies]

This is causing issues on Windows, which doesn't recognize np.float128.

AppVeyor output

[coveralls]

Coverage increased (+0.0003%) to 96.255% when pulling 0e49aee on matthew-brett:mac-precision-fix into 0d02553 on nipy:master.

[coveralls]

Coverage increased (+0.0003%) to 96.255% when pulling ee7183e on matthew-brett:mac-precision-fix into 0d02553 on nipy:master.

[effigies]

Looks good.