Fix image metadata being used in wrong cell

[flying-sheep]

Fix bug pointed out in #588 (review) caused by reusing the image_options dict. Fixes #605.

Seems like in my previous PR I didn’t understand the tests enough to realize that I changed them to nonsense, but now I do.

Sorry for that!

cc @OriolAbril

[flying-sheep]

I changed the metadata of the plot to reflect the text/plain representation (<Figure size 432x288 with 1 Axes>). This results in the plot looking blurry since the image data’s resolution is smaller (390×248). While it doesn’t look nice in the test notebook, it makes the most sense like that.

(The perfect solution would be to have quadruple-sized “retina” image data – e.g. 800×600 – with the metadata being half that – 400×300 – which would match real-world usage)

[OriolAbril]

Changes look good and I didn't notice anything strange after testing the PR locally. Thanks for the quick fix @flying-sheep!

[juanitorduz]

Hi! It would be great to have this fix into the next release 😄.

[agoose77]

Thanks for your feedback everyone! I am doing a round of maintenance on the Sphinx stack at the moment, I will aim for a release next week.

[flying-sheep]

Hi @agoose77 are you still planning to do a release soon?

[agoose77]

Thanks @flying-sheep, this is very helpful. Another contributor also opened a PR #609, so I'm in the fortunate position of having to make a choice between the two! As there's a new test in #609, I'm going to close this in favour of that one :)

[OriolAbril]

@agoose77 I would merge both PRs actually. They do have the addition of the .copy() in common, but that is all, so I would not expect merge conflicts either. This PR fixes the existing tests ehich are as of now testing that the wrong behaviour happens (consequently, all tests are failing on the other PR). The other PR doesn't change anything about existing tests (which needs to happen too in addition to fixing the bug) and adds a new test that is more specific for image metadata.

If you prefer, I can also merge both PRs into one keeping the commits of each author so contributions are properly tracked and open a new with the changes from both PRs

[agoose77]

@OriolAbril yep, I'm pulling the changes from the tests to attribute @flying-sheep :)