While testing we noticed some regressions, that were introduced with the commits 4f60cb6 and 7a04f7a. These commits were focused on providing a proper point coordinate system when multiple zoom values are involved and rescaling at runtime is active, but they also introduced behavioral changes when the rescaling at runtime is not active.

There are mainly three possible approaches to handle this regression:

1. Properly rework the coordinate system to support multi zoom environments. This is not doable for the upcoming release, but should be done for the next release
2. Only revert 4f60cb6 and 7a04f7a. This would solve the regression, but re-introduce some issues in the multi-zoom environment, when rescaling at runtime. As we will introduce the rescaling feature with the upcoming release via the experimental flag, this would negatively affect it.
3. Retain the adaptions added 4f60cb6 and 7a04f7a, but properly guard them to differentiate between the scenarios with rescaling at runtime active and inactive.

This PR uses the 3rd approach. The first commit is a squashed commit to revert 4f60cb6 and 7a04f7a. The second commit reintroduces the changes with guards.

One scenario to test

Primary Monitor: 125% (Left)
Secondary Monitor: 175% (Right)

When moving the workbench to the secondary monitor and opening the search window, the window appears on the primary (fallback) monitor instead of the secondary monitor.