Account for stacks in angle calculation

[Givikap120]

The stable version of "correct" angle calculation.
Made with doubles in mind since they're the main issues when it comes to angle-related bonuses.
Flashlight changes are not included intentionally because I'm not sure what is expected here.
PRing this for research and discussion purposes (in the sense that is it needed at all), also to calculate sheet of deltas.

Brief explanation of "why it's so complex":
It calculates 3 angles, where first angle is normal, second is accounting for one double, third is accounting for two doubles in a row. Each angle has weight of how likely that the respective objects are stacked.
Compared to "stack threshold" approach - this one is smooth and stable, meaning that as distance between notes in the stack would increase - it would be smoothly transitioning between "stacked" and "non-stacked" values, instead of big jump when distance passed certain threshold.

[stanriders]

Doubles have ~0 lowest velocity hence 0 angle bonus anyway, this is redundant for aim calculation

[Givikap120]

Doubles have ~0 lowest velocity hence 0 angle bonus anyway, this is redundant for aim calculation

For aim calculation - maybe (even tho there's deltas as high as -100pp). But in other places (reading for example) - you're forced to do some anti-doubles measures, because otherwise wrong angle can significantly screw up the calculations.