cargo-mutants VS rs-pbrt-test-scenes

Hi Rusty!

For a later hack in the same vein, check out https://github.com/sourcefrog/cargo-mutants

cargo-mutants got a couple of releases over the weekend, fixing some environment-triggered bugs on Windows and Linux ecryptfs, and adding a better overall progress indicator (demo: https://asciinema.org/a/481375). It is getting better at avoiding uninteresting or pointless mutations, although there's a lot more to do there, and at printing function name paths well.

While working on Conserve and cargo-mutants I felt that existing Rust progress-bar libraries like Indicatif didn't give me enough control of how the bar was drawn, or at least I was working around them to insert my own text. For example, I wanted to draw several counters into a single line, and it seems like the only way to do that is to just push in a text string.

Continuing with some fast-follow enhancements in cargo-mutants in response to user feedback: for example doctests are typically slow and in some trees not important for coverage, so you can now skip them.

I got cargo-mutants to a state where I felt I could announce it and ask for public feedback.

So, I started working on a library to deal with reading data from Blender‘s binary .blend files. In the past I was able to render some Blender v2.79 files, now I abstracted the library to deal (in theory) with all Blender versions (using the DNA information). I tested with two scenes for Blender 2.79 and Blender 3.0 (which should be quiet different from the GUI perspective). With v2.79 I got away with a naming convention and mis-using the old material settings (pre Cycles renderer). Blender 3.0 forced me to get some material settings from shader nodes and the library allows to follow pointers from one data structure to the next. Prove of concept: Both, the Cornell box, as well as the Radiance Conference Room, render with both Blender scene versions. Get the latest Blender scenes here: https://gitlab.com/jdb-walter/rs-pbrt-test-scenes/-/tree/master/blend