Make examples more deterministic for automated testing

[skalarproduktraum]

We are currently using the examples as integration tests as well. Some of them use randomised values, e.g. for lighting. As the screenshots created from these examples are automatically compared via Argos CI (see e.g. https://www.argos-ci.com/scenerygraphics/scenery/builds/38), this throws up quite a few false positives.

Ideas to fix this:

• use a fixed random seed in the example runs on CI to produce deterministic values
• don't use random values at all

[elect86]

Could you provide some link where those values are exactly set?

[skalarproduktraum]

An example would be https://github.com/scenerygraphics/scenery/blob/108ba6affef799633400ffb3e6080f685fdf8b75/src/test/kotlin/graphics/scenery/tests/examples/advanced/RibbonExampleSecondaryStructures.kt#L29

There's plenty of places in the examples where Random.random3DVectorFromRange() is used e.g. for light colors.

[elect86]

We might generate a static random palette offline and fetch during runtime from that

Or, we generate the palette at the start. If on CI, the palette will be always the same