empirical p-value error, maybe caused by numpy 1.10

[matthew-brett]

This just came up on travis:

======================================================================
ERROR: nipy.algorithms.statistics.tests.test_empirical_pvalue.test_efdr
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/nose/case.py", line 198, in runTest
    self.test(*self.arg)
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/nipy/algorithms/statistics/tests/test_empirical_pvalue.py", line 31, in test_efdr
    np.testing.assert_array_less(efdr.fdr(3.0), 0.2)
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/nipy/algorithms/statistics/empirical_pvalue.py", line 310, in fdr
    self.fdrcurve()
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/nipy/algorithms/statistics/empirical_pvalue.py", line 234, in fdrcurve
    self.learn()
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/nipy/algorithms/statistics/empirical_pvalue.py", line 206, in learn
    hist, ledge = np.histogram(x, bins=bins)
  File "/home/travis/build/matthew-brett/nipy/venv/lib/python3.5/site-packages/numpy/lib/function_base.py", line 249, in histogram
    n += np.bincount(indices, weights=tmp_w, minlength=bins).astype(ntype)
ValueError: operands could not be broadcast together with shapes (126,) (127,) (126,) 
----------------------------------------------------------------------

https://travis-ci.org/matthew-brett/nipy/jobs/84747582

I wonder whether this is an incompatibility with numpy 1.10 ? I seem to remember numpy changed something subtle with its histogram bin selection in this release.