StreamLines sets seed for np.random but not for random

Enhancement proposal

The streamlines are always rendered. The cached version is never used. The reason is that the code for StreamLines sets np.random.seed(0), the seed for np.random. However, another radomizer is actualy used. I have several sugestions:

• set the radomizer random.seed() as well, or
• use only the radomizer from numpy to generate streamlines, or
• mention the necessity to set the randomizer in documentation.

Additional comments

The following code is from example on manim webpage. Adding the second and third lines speeds up the second and other compilations by a large factor.

from manim import *
import random
random.seed(10)


class ContinuousMotion(Scene):
    def construct(self):
        func = lambda pos: np.sin(pos[0] / 2) * UR + np.cos(pos[1] / 2) * LEFT
        stream_lines = StreamLines(func, stroke_width=2, max_anchors_per_line=30)
        self.add(stream_lines)
        stream_lines.start_animation(warm_up=False, flow_speed=1.5)
        self.wait(stream_lines.virtual_time / stream_lines.flow_speed)

Start by locating the StreamLines implementation and the np.random.seed(0) call. Trace which randomizer generates the streamlines and how caching depends on deterministic output. Done means the selected seeding approach makes subsequent renders use the cached version, or the documentation clearly explains the required randomizer setup.