Fix delta emitter sample weights in the direct integrator

[JurriaanD]

Description

Fixes #1861

The direct illumination integrator computes the (MIS) weight of emitter samples as

Float mis = dr::select(ds.delta, Float(1.f), mis_weight(
                    ds.pdf * m_frac_lum, bsdf_pdf * m_frac_bsdf) * m_weight_lum);
result[active_e] += mis * bsdf_val * emitter_val;

where

• m_frac_lum = emitter_samples / (emitter_samples + bsdf_samples)
• m_frac_bsdf = bsdf_samples / (emitter_samples + bsdf_samples)
• m_weight_lum = 1 / emitter_samples

The weight of delta emitters is simply set to 1 so we are adding 1 * bsdf * Li to the output for each sample. The more samples you take the brighter the output will be.
My fix consists of moving the multiplication by m_weight_lum down one line so that it is also applied for delta emitter samples.

This bug is only present in the direct illumination integrator as far as I can tell.

Testing

I added a test that constructs a scene with

• A sphere with radius 1 at (0,0,0)
• A point light at (0,0,2)
• A camera at (0,0,5), looking down the z-axis

The scene is rendered with the direct integrator with emitter_samples=1 and emitter_samples=10. The test asserts that the color of the center pixel is the same in both renders.

Checklist

• My code follows the style guidelines of this project
• My changes generate no new warnings
• My code also compiles for cuda_* and llvm_* variants. If you can't test this, please leave below
• I have commented my code
• I have made corresponding changes to the documentation (N/A)
• I have added tests that prove my fix is effective or that my feature works
• I cleaned the commit history and removed any "Merge" commits
• I give permission that the Mitsuba 3 project may redistribute my contributions under the terms of its license