Sanity-check the sample names before committing to any data processing

[fruce-ki]

Description of feature

I propose that sample names should be checked for validity at the beginning of the run. The aim would be to look for common character errors that are not accepted in filenames by some of the tools, such as whitespaces, and give the user the chance to fix them before beginning any of the data work.

I just had a run crash (v3.14.0) after a whole week of processing, because picard was not ok with a space in the middle of some filenames. I do visually inspect the samplesheet metadata I receive, but things can be missed and finding out only so late in the run is very inefficient.