Also, this change fixes redundant checks the case when a monitor
does not exist. Now Ebitengine checks a monitor existence at the
initialization.
Closes#2145Closes#2795
initialMonitorByOS could return nil when a cursor was at an extreme
position like the bottom of the display. Apparently, a cursor position
could take an inclusive range of the monitor size.
This change fixes this issue by fixing the comparison.
Even if initialMonitorByOS returns nil, a fallback primary monitor
should be used, so this is not a critical issue.
Updates #2794
This change fixes these issues:
* `currentMouseLocation()` returned a position in the macOS native
coordinate. This means the Y axis is upward, while the Y axis is
downward in the GLFW coordinate. This change adjusts the Y position.
* `(*monitors).monitorFromGLFWMonitor` always returned nil at least
on macOS. This change replaces this with a new method
`(*monitors).primaryMonitor`.
Updates #807Closes#2794
Monitors can have different device scale factors, and in this case,
it doesn't make sense to use device-independent pixels as unit for
monitor positions and sizes.
Updates #2778
