Mitsuba Scene Description (MSD) -- a pythonic Mitsuba scene builder

[pixelsandpointers]

Dear all,

I have been trying to find a way to create a Pythonic API around Mitsuba's plugin system, which allows it to be programmed in Python without using the dictionary format. See #1388.

I wanted to contribute to Mitsuba for a while now, and perhaps this is something that may help others. I don't want to wait any longer and gather some initial feedback on what I came up with:
Mitsuba Scene Description

The Mitsuba Scene Description or short MSD scrapes the plugin pages of the official documentation and generates a plugin API based on the content. It has no runtime dependencies, other than Mitsuba (for transformations).

An example:

import mitsuba_scene_description as msd
import mitsuba as mi
mi.set_variant("llvm_ad_rgb")

diffuse = msd.SmoothDiffuseMaterial(reflectance=msd.RGB([0.8, 0.2, 0.2]))
ball = msd.Sphere(
    radius=1.0,
    bsdf=diffuse,
    to_world=msd.Transform().translate(0, 0, 3).scale(1.0, 1.0, 1.0),
)

cam = msd.PerspectivePinholeCamera(
    fov=45,
    to_world=msd.Transform().look_at(origin=[0, 1, -6], target=[0, 0, 0], up=[0, 1, 0]),
)
integrator = msd.PathTracer()
sun = msd.PointLightSource(
    to_world=msd.Transform().translate(3, 4, 2), intensity=msd.RGB([3, 3, 3])
)

scene = msd.Scene(
    integrator=integrator,
    sensor=cam,
    shapes={"ball": ball},
    emitters={"sun": sun},
)


mi_scene = mi.load_dict(scene.to_dict())
rndr = mi.render(mi_scene)
mi.util.write_bitmap("test.png", rndr)

This is the initial release, and I am very happy to receive feedback and advice on how to polish this further, so we can have a scene builder pattern based on Python. Right now, it is a bit rough around the edges, some generations might be duplicated, and I plan to do some post-processing with the tree-sitter API to remove duplicates altogether.

Let me know what you think and file issues. :)
Ben