[Bug]: NavigationToolbar2QT should use @Slot annotation

[bersbersbers]

Bug summary

https://doc.qt.io/qtforpython-6/tutorials/basictutorial/signals_and_slots.html#the-slot-class says

We recommend marking all methods used by signal connections with a @QtCore.Slot() decorator. Not doing causes run-time overhead due to the method being added to the QMetaObject when creating the connection. This is particularly important for QObject classes registered with QML, where missing decorators can introduce bugs.

NavigationToolbar2QT does not do that.

Code for reproduction

import os

from matplotlib.backends.backend_qt import NavigationToolbar2QT
from matplotlib.backends.backend_qtagg import FigureCanvasQTAgg
from PySide6.QtWidgets import QApplication

os.environ["QT_LOGGING_RULES"] = "qt.pyside.libpyside.warning=true"

app = QApplication()
canvas = FigureCanvasQTAgg()
NavigationToolbar2QT(canvas)

Actual outcome

qt.pyside.libpyside: Warning: Registering dynamic slot "home()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "back()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "forward()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "pan()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "zoom()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "configure_subplots()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "edit_parameters()" on NavigationToolbar2QT. Consider annotating with @Slot()
qt.pyside.libpyside: Warning: Registering dynamic slot "save_figure()" on NavigationToolbar2QT. Consider annotating with @Slot()

Expected outcome

Silence

Operating system

Windows 10 (also Ubuntu 23.10)

Matplotlib Version

3.8.0

Matplotlib Backend

QtAgg

Python version

Python 3.12.0

Jupyter version

irrelevant

Installation

pip

[QuLogic]

It could do that, but is that portable to PySide2 and PyQt? We would need compatibility with all Qt Python implementations, which may require lots of special casing.

[bersbersbers]

I will provide a PR shortly which can be used as a basis for such testing.