fleshing out check_numpy_array

[machow]

Copied from @beangoben 's issue raised here: datacamp/pythonwhat#217

(CC @yrdatacamp @hugobowne)

Feature request:

Having a test function catered to 1-D, 2-D numpy arrays, in particular that could check:

• If defined (already implemented)
• Have the same datatype (i,g list vs np.array)
• Have the same shape (i.g (2,3) vs (3,2))
• Has the same contents, or very close (like numpy's allclose https://docs.scipy.org/doc/numpy-dev/reference/generated/numpy.testing.assert_allclose.html), equality is only implemented

Currently if I wanted to do this I have to do the following test:

key = 'y'
# messages
type_msg="__JINJA__: **type** **mismatch**: expected `{{sol_eval}}`, got `{{stu_eval}}`."
type_expr = 'type({:s})'.format(key)
shape_msg="__JINJA__: **shape** **mismatch**: expected `{{sol_eval}}`, got `{{stu_eval}}`."
shape_expr = '{:s}.shape'.format(key)
contents_msg="__JINJA__: **solution** **contents** are different, expected `{{sol_eval}}`, got `{{stu_eval}}`."

Ex().check_object(key)  \
    .has_equal_value(expr_code = type_expr,incorrect_msg=type_msg)  \
    .has_equal_value(expr_code = shape_expr,incorrect_msg=shape_msg) \
    .has_equal_value(incorrect_msg=contents_msg)

Would nice to have a function that could wrap around all these scenarios such as

test_numpyarray(key,type_test=True,type_msg,shape_test=True,shape_msg,content_test='allclose', contents_msg)