pyqtgraph as backend for 2D-Data-Browser

In my 2021 Google Summer of Code project I worked on providing an additional backend for the 2D-Data-Browser currently implemented with matplotlib for Raw.plot(), Epochs.plot() and ICA.plot_sources(). You can find the final report here
This is the PR which contains the new backend.

In this issue I will host an updated list with the tasks left to be done and ideas for future features in the new pyqtgraph-backend.
I would be much interested in feedback about the desing & features of the new backend and/or new ideas for things you would like to see in the new backend. If you got some, leave them in the comments.

ToDo for #9687

Tests

• make all tests in mne.viz.tests.test_raw.py pass for pyqtgraph

Features

• Event Lines
• Scale-Bars/Legends
• Apply projections with GUI
• Channel-Clipping
• Settings-Dialog with pyqtgraph-specific settings

Improvements (from feedback)

• Crosshair without horizontal line but dot

Other features from MPL

• adaption to Epochs.plot() (+ test_epochs.py)
• adaption to ICA.plot_sources() (+ test_ica.py)
• Per-channel-annotation (for Epochs)
• Selection-Figure (group_by-parameter)

New feature ideas

• Right-Click Dragging for scaling
• Pinching for zooming with trackpad
• custom colors for annotations (like in event-colors, needs to be adapted in mne-python)
• Dark Mode (preferably with parameter accepting dict for colors to be adjusted by user)
• Filter view-range dynamically (set filter-parameters in gui)
• Initialize Browser without data (and load data from menu/with drag&drop)
• Start Data-Browser from command-line