Related to #4353, #4980, but would like to ask the cases when the indexer is a single bool.

1. In normal array, bool looks coerces to int. Is this intended, or should be regarded as a boolean indexing with different shape (DEP: Deprecate boolean array indices with non-matching shape #4353)?

   np.__version__
   # '1.9.2'
   
   np.array([0])[True]
   # IndexError: index 1 is out of bounds for axis 0 with size 1
   np.array([0])[False]
   #0
   

2. 0-dim array behaves quite strangely. I understand this should raise IndexError if it coerces to int.

   np.array(0)[False]
   # array([], dtype=int64)
   np.array(0)[True]
   # array([0])
   
   np.array(0)[0]
   # IndexError: too many indices for array