[FEATURE]: Make samples explicit and brief speed guide

**Is your feature request related to a problem? Please describe.**
Required samples are highly scene dependent. We have hidden default values that often take unnecessarily long.

**Describe the solution you'd like**
We should make the samples explicit in examples again so that users have guidelines and the option to iterate faster.

**Describe alternatives you've considered**
A quickmode option. Or low default amount of samples, but this might lead to smoothed high-frequency artifacts of the denoiser depending on the scene. We don't want to assume that this is okay for every learning task even though it is for some.