feat: resize & focus tracking in view_data

[iisakkirotko]

Pull Request Template

Description

Stores timestamps resized_at and focused_at in view_data. This is useful to track which viewer figures should have their shape set based on.

@maartenbreddels and others I welcome discussion / feedback on the focus portion in particular:

• Should we also track the last point at which focus entered a view?
• Is there any need to remove the eventListener from window?

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 10 lines in your changes missing coverage. Please review.

Project coverage is 42.42%. Comparing base (7b90b9d) to head (21609d0).
Report is 16 commits behind head on main.

Files with missing lines	Patch %	Lines
bqplot_image_gl/viewlistener.py	0.00%	10 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #112      +/-   ##
==========================================
- Coverage   50.00%   42.42%   -7.58%     
==========================================
  Files           7        8       +1     
  Lines         140      198      +58     
==========================================
+ Hits           70       84      +14     
- Misses         70      114      +44     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[maartenbreddels]

Nice PR!
The only worry I have is that a particular view could also have received the focus event. We could also implement this later on, so we don't just listen to the window focus event, but also the element's focus event.
What do you think?

[mariobuikhuizen]

Why are we storing all views if we only use the last updated view?

[maartenbreddels]

That is an implementation detail (or particular use case) of glue-jupyter, this is a general widget.

[iisakkirotko]

So I think I managed to track down the issue, which was in a different place than I thought. Looks like the tox scripts that are used to run tests (OpenAstronomy/github-actions-workflows/.github/workflows/tox.yml@v1) seem to be using conda for installing dependencies. However, the conda-forge package for typing_extensions is maintained by a third party (see this comment), and claims to support Python 3.7 even in current versions, while actually support for Python 3.7 was dropped in 4.8.0. This causes conda to resolve an actually incompatible version of the package, which in turn causes the tests to fail. How to solve it... I'll look into :)

[maartenbreddels]

Thanks, also a nice split in commits