Improve manual implementations of up- and downcasting

[jnbooth]

This PR provides some minor improvements to #1159 for the implementations of Upcast.

• Replaces downcastPtr's use of dynamic_cast with qobject_cast, which is much faster and does not require RTTI.
• Implements Upcast<QVector<QPoint>> for QPolygon, replacing the standalone functions previously used for Deref and DerefMut.
• Implements Upcast<QVector<QPointF>> for QPolygonF, replacing the standalone functions previously used for Deref and DerefMut.
• Implements Upcast<QCoreApplication> for QGuiApplication and adds Deref. ¹
• Implements Upcast<QGuiApplication> for QApplication and adds Deref. ¹
• Implements Deref for QQmlApplicationEngine to QQmlEngine using the existing upcasting implementation. ¹
• Removes a null pointer created inside QStringList's implementation of Upcast<QList<QString>> due to a misuse of dynamic_cast.

¹ Removed in favor of #1202.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (fb83a89) to head (1baaae1).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #1204   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           73        73           
  Lines        12681     12681           
=========================================
  Hits         12681     12681           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[LeonMatthesKDAB]

@jnbooth Thank you for the contribution, this is looking good.

Regarding the downcasting, I think we can get all those concerns implemented in a sane way with a bit of meta-programming that automatically selects between qobject_cast and dynamic_cast.

And keeping static_cast for the special cases is okay, though I think we should add another assertion just out of caution.

[jnbooth]

@LeonMatthesKDAB I don't know how to do that kind of metaprogramming. Are there examples of it in the codebase that I could learn from? Alternatively, should I strip down part of this PR so it can be merged without that?

[LeonMatthesKDAB]

@jnbooth This should do it:
LeonMatthesKDAB@fdd65e9
In case you're interested, google SFINAE and enable_if_t, it's one of the crazier C++ concepts for sure.
But is quite powerful for cases like this.

I pushed this to a branch on my fork (non-qobject-casting) and rebased, feel free to cherry-pick or just use that branch for your PR.

My template meta-programming is a bit rusty nowadays, so @ahayzen-kdab please review carefully to check this is sane.

[jnbooth]

Any updates on this? I think at minimum it would be good to remove the null pointer returns.

[jnbooth]

I don't have approve permissions btw, so it is up to someone else to do that.

[LeonMatthesKDAB]

Ah sorry @jnbooth this PR probably lost the race with another approval which outdated the auto-merge, this is an annoying issue with our workflow at the moment 🙄
It should auto-merge now after the CI passes 🤞