Fix audio panning

[Speedy-Consoles]

Closes #1939.
The big mistake seems to be using Dist instead of dx in

if(dx < 0)
	a -= (Dist/m_MaxDistance)/2.0f;
else
	a += (Dist/m_MaxDistance)/2.0f;

I also don't see a reason to use the sine here.

[msiglreith]

The big mistake seems to be using Dist instead of dx in

yeap, my bad.

I also don't see a reason to use the sine here.

Yield those graphs, which should result in a constant perceived sound volume (squares of gains) independent of the position (pan more to left or right speaker) excluding the linear distance based falloff

[Speedy-Consoles]

I see, but shouldn't we be using square-law panning instead of sine-law panning, so the "perceived sound volume" of each channel is still linear with dx? I think sine-law panning would make sense, if we were using an angle as pan value, but we're using a distance.

[msiglreith]

I'm not aware of an additional property which would justify sqrt or sin/cos over the other. I guess it boils down to preference of the listener here, but I'm no audio engineer so I'm fine with both.

[Speedy-Consoles]

I'm no audio engineer either, but i'm pretty sure it should be the square root instead. The difference in both would be, whether your pan value is proportional to the angle or the x coordinate of the sound source. The latter is the case here.

[teetow]

Quick comment: Our design was that on-screen sounds should play more or less centered, with a "dead zone" around the player that should roughly cover the screen. As soon as a source leaves the screen, th panning should become noticeable.

Dunno what has been done to the game since, but I always thought that design made sense.