Remove use of matplotlib's get_offset_position

[NeilGirdhar]

Code PR

• I have read through the contributing notes and understand the structure of the package. In particular, if my PR modifies code of plotly.graph_objects, my modifications concern the codegen files and not generated files.
• I have added tests (if submitting a new feature or correcting a bug) or
  modified existing tests.
• For a new feature, I have added documentation examples in an existing or
  new tutorial notebook (please see the doc checklist as well).
• I have added a CHANGELOG entry if fixing/changing/adding anything substantial.
• For a new feature or a change in behaviour, I have updated the relevant docstrings in the code to describe the feature or behaviour (please see the doc checklist as well).

[NeilGirdhar]

Fixes #3624 ?

[alexcjohnson]

Does this restore compatibility all the way up to the current MPL v3.8.0? Kind of remarkable that code written 9 years ago, for MPL v1.3, kept working through two major version bumps but broke on a minor 😅

In addition to my comment above, there are two things we'll want to do before merging:

• Add a changelog entry (make a new ## UNRELEASED -> ### Fixed section at the top)
• The test requirements files requirements_*_optional.txt all include MPL v2.2.3 (only 5 years old 🙈) - perhaps we can leave this version in requirements_36_optional.txt to ensure backward compatibility but bump 37, 38, 39, and 39_pandas_2 to the latest supported by each Python version so the existing matplotlylib tests run on the versions new users will get by default. Looks like that's:
  • Py3.7: MPL v3.5.3
  • Py3.8: MPL v3.7.3
  • Py3.9: MPL v3.8.0

[NeilGirdhar]

@alexcjohnson Done, as requested.

for MPL v1.3, kept working through two major version bumps but broke on a minor 😅

Good question. Maybe it was deprecated and no one noticed?

Thanks for your quick attention to this issue. It totally breaks WandB's use of plots, so it will be awesome to get this in.

[alexcjohnson]

oh dear, may have opened a can of worms upgrading those dependencies, here's the failure from Py3.9:

The conflict is caused by:
    The user requested numpy==1.20.2
    pandas 1.2.4 depends on numpy>=1.16.5
    xarray 0.17.0 depends on numpy>=1.15
    ipython[all] 7.22.0 depends on numpy>=1.16; extra == "all"
    scipy 1.6.2 depends on numpy<1.23.0 and >=1.16.5
    matplotlib 3.8.0 depends on numpy<2 and >=1.21

Are you able to see these logs @NeilGirdhar? I still think we should make these upgrades, since otherwise our tests are way out-of-date vs what users will actually install today. If you're able to sort these out (along with running black on the new code to fix the code_formatting error) that would be fantastic. If not, I can look into it but might take me a couple of days to get to it.

[NeilGirdhar]

Why don't we bump everything to at least NEP 29's defaults?

[NeilGirdhar]

@alexcjohnson Is there anything else I should do to polish this off?

[alexcjohnson]

💃 Nice work @NeilGirdhar! Merging, I expect we'll release this next week with the next plotly.js minor. Thanks!

[NeilGirdhar]

Thanks for the quick review process!

[NeilGirdhar]

@alexcjohnson Any chance we could get a release soon with this since it breaks WandB?

[alexcjohnson]

Sorry for the delay @NeilGirdhar - we’re just finishing up a feature in plotly.js and then will make a release. Should be the next couple of days.

[NeilGirdhar]

@alexcjohnson No problem at all! Looking forward to the release :)