Fix errors caused in jinja StrictUndefined mode

[pamelafox]

Fixes #2522

This PR addresses errors that arise when running jinja in StrictUndefined mode (like mypy for Jinja!). There were different ways that I addressed the issues depending on the attribute:

• extra_css/extra_js: I added these as attributes to the BaseView, since their usage and documentation implies that they should always exist as empty lists.
• form.csrf_token: This is only defined if the dev is using SecureForm as the base class, which isn't always the case, so I used the pattern of "form.csrf_token is defined and form.csrf_token" which seems to be the recommended way in Jinja to ensure a variable exists before using it.
• return_url: From what I could tell, this is only passed in sometimes? So I made an if to only use it if it exists. But I might be wrong about that.
• widget.input_type: Only the classes that output HTML input fields have this defined. I considered adding input_type: Optional[str] = None to all the other classes, but that would have been a larger change, so I added the "is defined" check to templates instead.
• config.FLASK_ADMIN*: Since these will only be defined by devs who want to use that particular functionality, I used the "is defined" pattern for them.

Generally, for most cases, I ended up using the "is defined" pattern, but if you think there are more situations where .py changes would be warranted, let me know!

[pamelafox]

I wasn't sure if 1) this change warranted an entry,
2) this was the correct version header

[samuelhwilliams]

This looks great - that'll be the next version, so perfect :)

[pamelafox]

This is not the optimal import sorting, given that "admin" is a local folder, but ruff does not know that. I could use "." to make it clearer, or we could just be okay with it.
This seems to be the only example complex enough to have both an app.py and a folder of other modules.

[samuelhwilliams]

I am personally fine with it. If someone was so inclined, we could probably add some ruff config that would help it sort "correctly" - but I consider it broadly unimportant given the small scope, as you note.

[samuelhwilliams]

Great work on this - thank you :)