Slow as Hecks on OSX

knnimpute is slow on OSX because of missing ATLAS/BLAS. Confirm with:

import numpy.distutils.system_info as sysinfo
sysinfo.get_info('atlas')
np.__config__.show()

-> import numpy.distutils.system_info as sysinfo
(Pdb) import numpy.distutils.system_info as sysinfo
(Pdb) sysinfo.get_info('atlas')
{}
(Pdb) np.__config__.show()
lapack_opt_info:
    libraries = ['openblas', 'openblas']
    library_dirs = ['/usr/local/lib']
    define_macros = [('HAVE_CBLAS', None)]
    language = c
blas_opt_info:
    libraries = ['openblas', 'openblas']
    library_dirs = ['/usr/local/lib']
    define_macros = [('HAVE_CBLAS', None)]
    language = c
openblas_info:
    libraries = ['openblas', 'openblas']
    library_dirs = ['/usr/local/lib']
    define_macros = [('HAVE_CBLAS', None)]
    language = c
blis_info:
  NOT AVAILABLE
openblas_lapack_info:
    libraries = ['openblas', 'openblas']
    library_dirs = ['/usr/local/lib']
    define_macros = [('HAVE_CBLAS', None)]
    language = c
lapack_mkl_info:
  NOT AVAILABLE
blas_mkl_info:
  NOT AVAILABLE

Related:
https://github.com/CellCognition/cecog/wiki/Build-build-accelerated-numpy-using-ATLAS-on-Mac-OSX
https://joernhees.de/blog/2013/06/08/mac-os-x-10-8-scientific-python-with-homebrew/

Start by reproducing the reported diagnostics on OSX: run the numpy.distutils atlas check and np.config.show() in the project environment. The issue provides no file or test entry point; done would mean determining how the missing ATLAS/BLAS support affects knnimpute and defining a verified resolution for the reported slowdown.