(Following a02c2d0c.)

This includes two minor bugfixes related to proportional PDFs from BxDFs and
volumetric emission that also eliminate consistencies with the VolPathIntegrator.

s/T_hat/beta/g
s/uniPathPDF/inv_w_u/g
s/lightPathPDF/inv_w_l/g

In preparation for changes corresponding to a02c2d0c. Note that
the variable names do not match their actual usage, but this makes the forthcoming
actual logic changes easier to review.
logic changes

Variables were removed or renamed in 59b5988d and 24c6ef53

Qualify Medium::IsEmissive() with PBRT_CPU_GPU

Get rid of the MediumSample structure; pass the various appropriate values
directly to the lambda passed to SampleT_maj() rather than wrapping them
up into that.

Simplify MediumInteraction to not include sigma_* and Le (it's not necessary
and just causes a bunch of data copying.)

Make SampleT_maj a standalone function
Otherwise minor variable renaming and restructuring.

Rather than tracking them along with the ray (and incurring the
corresponding bandwidth and memory costs), we always use the
Camera::Approximate_dp_dxy() method, which may mis-estimate them in the
presence of specular reflection and refraction, though generally works
well.

Mostly addresses issue #10.

This way, we can not increment the ray depth when it passes through a
surface that markes the boundary between different media, matching the
behavior of the regular CPU path.

In particular, this fixes some undesirable noise in the wavefront
integrator (and different behavior from the regular CPU path) that was due
to the ray depth being incorrectly tracked in scenes with participating
media and then Russian roulette inadvertently being applied after the very
first surface intersection.

* Generalize GPU rendering path to run on both CPU and GPU

Now we have a WavefrontIntegrator that can both run on the CPU (backed by
ParallelFor() for parallelization and pbrt's aggregates for ray
intersection acceleration) and on the GPU (backed by GPUParallelFor() for
parallelization and OptiX for ray intersection on NVIDIA GPUs.)

Beyond generalizing the code, this refactor allows CPU-side debugging and
testing of the wavefront integrator.  Doing so allows further isolation of
the GPU-specific code into a few source files, now just ~2.5k lines of code.

This includes a bug fix in the wavefront medium code to resolve MixMaterials
to one of their constituent materials before enqueuing material evaluation
and shading work.

Note that on the CPU, the wavefront integrator runs 5-10x more slowly than
pbrt's regular CPU integrators, so it is not recommended for regular use...
Co-authored-by: NWenzel Jakob <wenzel.jakob@epfl.ch>