Fix sensitivity of tangent editing in Automation Editor

[regulus79]

Previously, the sensitivity of the tangent editing in the automation editor was hardcoded so that it would change the internal value by the number of pixels the mouse moved up (with a bit of scaling based on the x position too). This is not correct, since it does not respect the zoom of the editor or the min/max values relative to the window size.

This PR adds the correct scaling constant to the tanget value to convert pixels to levels. The automation editor already had the right constant as a member variable, m_y_delta, but it would only update when using the hardcoded zoom levels, so I reworked it to also work for auto zoom.

[szeli1]

I did not test this PR, so I'm not approving it, but the code changes look okay.

[allejok96]

Very minor issue: I noticed that the little handle still doesn't quite line up with the mouse. When dragging it in an empty not-connected automation clip, it lines up perfectly with the mouse. But on a clip connected to the volume knob I have to hold the cursor 8 pixels to the right of the dot to have it line up in the Y direction.

[regulus79]

Very minor issue: I noticed that the little handle still doesn't quite line up with the mouse. When dragging it in an empty not-connected automation clip, it lines up perfectly with the mouse. But on a clip connected to the volume knob I have to hold the cursor 8 pixels to the right of the dot to have it line up in the Y direction.

I can also sort of reproduce this. I assume it's due to imprecisions since m_y_delta is an int. But now I've changed it to be a float, so hopefully it's fixed in the most recent commit?

[allejok96]

Works like a charm now!

[messmerd]

Haven't tested, but looks pretty good

[messmerd]

Not part of this PR, but just below this there's this:

const QString & zfac = m_zoomingYModel.currentText();
m_y_auto = zfac == "Auto";

The m_zoomingYModel's first item, "Auto", is not a QObject::tr() translated string, but it should be. However, if there were any translation that was not "Auto", this code here would break. This is all very finicky. I think we should be checking the current index not the current text, so:

m_y_auto = m_zoomingYModel.value() == 0; // "Auto"

[allejok96]

Nice catch, also I fixed this in #7941