feat: deprecation warnings on select application attributes

[dimaqq]

Description

• explicitly export attributes accessible on Application via properties
• mark some as deprecated, as these are unlikely to be available in Juju 4

[james-garner-canonical]

I had a couple of questions that I accidentally left "pending" all weekend, but thankfully Ben asked them today. Do you think it would be worth describing anything about Application's properties in the class docstring?

[dimaqq]

This package already imports typing_extensions via some dependency.

I've made it an explicit dep, allowing for a simpler @deprecated decorator.

Here's a test run, showing correct source for the warning:

/code/python-libjuju/example.py:28: DeprecationWarning: Application.owner_tag is deprecated and will be removed in v4
  owner_tag.......... {app.owner_tag!r}
/code/python-libjuju/example.py:30: DeprecationWarning: Application.min_units is deprecated and will be removed in v4
  min_units.......... {app.min_units!r}
/code/python-libjuju/example.py:32: DeprecationWarning: Application.subordinate is deprecated and will be removed in v4
  subordinate........ {app.subordinate!r}
/code/python-libjuju/example.py:34: DeprecationWarning: Application.workload_version is deprecated and will be removed in v4, use Unit.workload_version instead.
  workload_version... {app.workload_version!r}

[dimaqq]

I see doc changes as orthogonal, so maybe in a separate PR?

[dimaqq]

CI failures are unrelated, #1108

[james-garner-canonical]

The @deprecated decorator is cool, didn't know about that. One or two result in rather long lines, but I think we can look at that later if we introduce linting with ruff/etc.

The CI failures are all known failing tests. 3/5 fail 100% of the time on main.
test_wait_for_idle_more_units_than_needed fails 80% of the time on main.
test_subordinate_false_field_exists fails around 25% of the time on main.

The changes made in this PR seem reasonable to me, but I would like the source of truth for the new property names and their type annotations to be mentioned somewhere, if not in a docstring then perhaps somewhere it will show up in the git log.

"""Class representing a juju application.

Explicit properties are annotated with data types inferred from ...
See ... juju documentation page?
"""

I think I want this because I don't know where they come from or why they're correct, and maybe a future reader will wonder the same thing. But if you don't think this is worthwhile here then let's skip it for now.

Also, maybe we want a <5 restriction on the typing_extensions dependency just to be conservative? I know the existing dependencies are loosely specified, but that already caused at least one CI breakage.

[dimaqq]

The changes made in this PR seem reasonable to me, but I would like the source of truth for the new property names...

Done, ptal

[dimaqq]

/build

[dimaqq]

/merge

[dimaqq]

/merge