Fix NumPy deprecation warning

[jgosmann]

Motivation and context:

See 1 for details in the release notes. Using np.floating will
capture more types than np.float, but is arguably closer to what one
would expect from an is_float function and consistent with the
is_integer function.

Interactions with other PRs:

none

How has this been tested?

existing tests

How long should this take to review?

• Quick (less than 40 lines changed or changes are straightforward)

Types of changes:

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• I have read the CONTRIBUTING.rst document.
• [n/a] I have updated the documentation accordingly.
• I have included a changelog entry.
• I have added tests to cover my changes.
• I have run the test suite locally and all tests passed.

I think this change does not need a changelog entry as it shouldn't affect the public API, but if desired I can add one.

[tbekolay]

I think np.floating is not meant to be used this way; a few things I tried in the terminal:

>>> isinstance(1.0, np.floating)
False
>>> isinstance(np.array(1.0), np.floating)
False
>>> isinstance(np.array([1.0]), np.floating)
False

Probably to make this work we'll need to check if it is indeed a NumPy array first, then check the dtype; e.g.

>>> np.issubdtype(np.array([1.0]).dtype, np.floating)
True

This might do the same thing, and seems a bit more obvious:

>>> np.array([1.0]).dtype.kind == "f"
True

Edit: This might mean that our is_int implementation is wrong and should be fixed as well, I didn't look back at it.

[jgosmann]

I updated the PR. This probably needs to be fixed in Nengo core too.

[jgosmann]

Finally got a green build. :)