[EXPERIMENTAL] Improve rendering time by rewriting STD140BufferFormat._write_padded()

[chopan050]

The original implementation of STD140BufferFormat._write_padded() used np.pad which is slow. In the given example scene, this method takes almost half of the rendering time.

This new implementation avoids that by creating a new array of zeroes instead. In this way, this method now takes 60x less time, and the total rendering time is almost halved.

Example scene:

class Twist(ThreeDScene):
    def construct(self):
        self.set_camera_orientation(theta=TAU/8, phi=TAU/5)

        def function(u, v):
            x = 8*u - 4
            y = 2*v - 1
            z = 1.0
            angle = TAU/4 * u
            point = (
                np.cos(angle) * np.array([x, y, z])
                + np.sin(angle) * np.array([x, -z, y])
            )
            return point
        
        surface = VGroup(Surface(function, resolution=(20, 5)))
        for i in range(3):
            face_copy = surface[0].copy()
            face_copy.rotate(TAU/4 * (i+1), RIGHT, ORIGIN)
            surface.add(face_copy)

        self.add(surface)

        self.begin_ambient_camera_rotation()
        self.wait(5)

Reviewer Checklist

• The PR title is descriptive enough for the changelog, and the PR is labeled correctly
• If applicable: newly added non-private functions and classes have a docstring including a short summary and a PARAMETERS section
• If applicable: newly added functions and classes are tested