RF: Use cython imports instead of relying on extern

[MarcCote]

Small refactoring as suggested by @matthew-brett (see comment #1076 (comment))

[codecov-io]

Codecov Report

Merging #1228 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #1228   +/-   ##
=======================================
  Coverage   85.88%   85.88%           
=======================================
  Files         221      221           
  Lines       27285    27285           
  Branches     2785     2785           
=======================================
  Hits        23435    23435           
  Misses       3165     3165           
  Partials      685      685

---

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 568d14c...287d30e. Read the comment docs.

[matthew-brett]

Finishing up stdlib import removals at MarcCote#10 .

[coveralls]

Coverage remained the same at 88.391% when pulling 287d30e on MarcCote:rf_use_libc_stdlib into 568d14c on nipy:master.

[MarcCote]

@matthew-brett thanks. @arokem this PR is ready.

[arokem]

Darn. Should've probably asked you to rebase on master before merging. Looks like this broke the master branch: https://travis-ci.org/nipy/dipy/jobs/224600749. Any idea how these things interact?

[matthew-brett]

Error is:

======================================================================
FAIL: dipy.reconst.tests.test_dki_micro.test_single_fiber_model
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/build/nipy/dipy/venv/lib/python2.7/site-packages/nose/case.py", line 197, in runTest
    self.test(*self.arg)
  File "/home/travis/build/nipy/dipy/venv/lib/python2.7/site-packages/dipy/reconst/tests/test_dki_micro.py", line 131, in test_single_fiber_model
    assert_almost_equal(wmtiF.tortuosity, ADe/RDe)
  File "/home/travis/build/nipy/dipy/venv/lib/python2.7/site-packages/numpy/testing/utils.py", line 468, in assert_almost_equal
    raise AssertionError(msg)
AssertionError: 
Arrays are not almost equal to 7 decimals
 ACTUAL: 2.59770124494259
 DESIRED: 2.597701149425287

I immediately can't think of any way the stdlib imports could cause a change at at the 7th decimal place, only for Cython 0.25.1 / numpy 1.7.1. Maybe there's some randomness in the test?

[arokem]

You're right. This is also relatively new, so might just be the precision wasn't set appropriately: #1231

To answer your question: yes, there is some randomness in that test: https://github.com/nipy/dipy/blob/master/dipy/reconst/tests/test_dki_micro.py#L88-L89