[rshapes] More spline functions

[AmityWilder]

Description

It would be helpful to have spline functions for velocity--to sample a curve's direction--and calculating bounds--for collision/rendering optimization. I think these would fit in nicely alongside the GetSplinePoint_ functions.

GetSplineVelocityLinear
GetSplineVelocityBezierQuad
GetSplineVelocityBezierCubic
GetSplineAccelerationBezierQuad
GetSplineAccelerationBezierCubic
GetSplineJoltBezierCubic

GetSplineBoundsBezierLinear
GetSplineBoundsBezierQuad
GetSplineBoundsBezierCubic

GetSplineCurvatureBezierCubic
GetSplineNearestTLinear

It might also be nice to have spline drawing with variable thickness along the curve, since this can't currently be accomplished without some pretty advanced trickery.

I have already written the code for these as part of my Raylib vector art program and thought they might be valuable for others as a contribution to Raylib.
The variable thickness one needs a little work because it doesn't translate 1:1 from object-oriented Rust directly into C (I will continue to work on that further), but it is already working pretty well as a basic calligraphic effect.

[AmityWilder]

Variable thickness sampling is now working as intended, as of commit 1e6332e

https://www.desmos.com/calculator/hswsmzuflh

[raysan5]

@AmityWilder Those improvements are really nice but I'm afraid they are growing considerably, maybe they should be moved to a separate library... let me review them carefully...

[AmityWilder]

@raysan5 If you think it would be better, I could split this issue into two separate ones: one for the math functions and one for the draw functions. Alternatively, I could split the draw functions off entirely and move them to raylib-extras, or create my own entirely separate library for them.

I believe the math functions (GetSpline_) are pretty much complete (I just need to make sure the bounds functions translated over accurately from Rust, since my original implementation used Rust iterator methods, and in the Raylib version I inlined certain functions I don't think would be used frequently enough to put into Raylib), but the variable-width draw functions may need a bit more work before they're ready for merging, since my Rust implementation of the variable width was reliant on a special data structure,

// amyvec::bezier::cubic

/// A cubic Bezier slice built from a [`curve::Slices`] iterator
pub struct Cubic {
  pub p0:    Vector2,
  pub c_in:  Vector2,
  pub c_out: Vector2,
  pub p1:    Vector2,
}

which I don't think makes sense to port into Raylib. I'm in the process of rewriting the functions to use an array [p0, c1, c2, p3, c4, c5, p6, ...] instead.

I will say however, in case it influences your decision, I'm thinking of further modifying the variable width draw functions so that Raylib users can specify the t-values of each thickness (e.g. [t0, thick0, t1, thick1, ...]), rather than simply having them evenly spaced along the spline.

[raysan5]

@AmityWilder As commented in the open PR, I think that moving the points-generation math and triangles-generation to its own single-file header-only library would be great, so rshapes module could include that library to implement the spline-drawing functions but the header would also be available to users with more advance spline-generation/management requirements. That library would also be platform/engine agnostic and portable.

I'm closing this issue but please, let's continue the discussion in the open PR: #4810