Check for logdet and numpy issues in python3

[mnarayan]

In applying AdaptiveGraphLasso to some new data, I see a lot of these warnings

/share/sw/free/anaconda/4.2.0/python3.5/lib/python3.5/site-packages/
numpy/linalg/linalg.py:1757:
RuntimeWarning: invalid value encountered in slogdet
  sign, logdet = _umath_linalg.slogdet(a, signature=signature)

/path/to/home/.local/lib/python3.5/site-packages/skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/metrics.py:25: 
RuntimeWarning: invalid value encountered in multiply
  fast_logdet(precision) - dim * np.log(2 * np.pi)

/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/metrics.py:116: 
RuntimeWarning: invalid value encountered in greater
  mask = np.abs(precision.flat) > np.finfo(precision.dtype).eps
/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/metrics.py:113:
RuntimeWarning: invalid value encountered in multiply
  l_theta = -np.sum(covariance * precision) + fast_logdet(precision)
/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/metrics.py:114: 
RuntimeWarning: overflow encountered in double_scalars
  l_theta *= n_features / 2.
/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/inverse_covariance.py:331: 
RuntimeWarning: invalid value encountered in less
  min_indices = np.where(np.abs(ebic_scores - ebic_scores.min()) < 1e-10)

  File "/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/quic_graph_lasso.py", line 888, in fit
    best_lam_idx = self.ebic_select(gamma=self.gamma)
  File "/path/to/home/.local/lib/python3.5/site-packages/
skggm-0.2.6-py3.5-linux-x86_64.egg/inverse_covariance/inverse_covariance.py", line 332, in ebic_select
    return np.max(min_indices)
  File "/share/sw/free/anaconda/4.2.0/python3.5/lib/python3.5/
site-packages/numpy/core/fromnumeric.py", line 2252, in amax
    out=out, **kwargs)
  File "/share/sw/free/anaconda/4.2.0/python3.5/lib/python3.5/
site-packages/numpy/core/_methods.py", line 26, in _amax
    return umr_maximum(a, axis, None, out, keepdims)
ValueError: zero-size array to reduction operation maximum which has no identity

Encountered this when using latest develop branch, e16ff8c4d10
and calling skggm with python3.5