Fix capitol image

[homosapien-lcy]

After the capitol.jpg moved, chaco/examples/user_guide/plot_types/create_plot_snapshots is broken, use importlib.resources to find the image for loading to fix this

[homosapien-lcy]

closes #879

[corranwebster]

This breaks on 3.7 and 3.8.

See https://github.com/enthought/pyface/blob/0fb8373b4dbb5e104d3120e88590dbf964865245/pyface/tests/test_pil_image.py#L14-L20 for an example of how to handle this sort of use of importlib across many Python versions

[corranwebster]

resources.files is only available on Python 3.9 or later, so this will break on 3.7 and 3.8 which we still want to support.

The fix is to use importlib_resources package which provides the newer API on older versions of Python

[homosapien-lcy]

Thanks for the tip, just add both importlibs

[dpinte]

LGTM

[corranwebster]

Typically you want to have this in a with statement something like:

resource = files(...)
with as_file(resource) as filename:
    ...

[corranwebster]

Can you please add an appropriate dependency to importlib_resources for Python < 3.9 here: https://github.com/enthought/chaco/blob/main/chaco/__init__.py#L19

Also a usage issue for importlib.resources.files.

[homosapien-lcy]

Can you please add an appropriate dependency to importlib_resources for Python < 3.9 here: https://github.com/enthought/chaco/blob/main/chaco/__init__.py#L19

Also a usage issue for importlib.resources.files.

Got it, just add a "check python3.9 then install importlib_resources" there.
Just want to clarify, what do you mean by usage issue for importlib.resources.files?

[corranwebster]

Just want to clarify, what do you mean by usage issue for importlib.resources.files?

Meaning it should be used in a with statement, which you have now done.

This should be good to merge.

[homosapien-lcy]

Just want to clarify, what do you mean by usage issue for importlib.resources.files?

Meaning it should be used in a with statement, which you have now done.

This should be good to merge.

Thanks Corran! The merge is still block by the ci test, the good old RuntimeError Numpy... Is there anyway we can get around it?

RuntimeError: NumPy was built with baseline optimizations: 
(SSE SSE2 SSE3 SSSE3 SSE41 POPCNT SSE42 AVX F16C AVX2) but your machine doesn't support:
(AVX2).