full/availability: Vet parameters in reconstruction tests

[Wondertan]

#2697 reworked our reconstruction tests to involve worst-case data withholding scenarios. This required us to dive deeper into the k, c, and s parameters in the original DA paper. As per @distractedm1nd, the c parameter does not represent the withholding case, so he made a custom script that simulates reconstruction and reevaluates c for withholding scenario.

However, we need more confidence in the revaluation results and to vet and analyze them more.

cc @musalbas

[walldiss]

Values for selected predefined parameters needs to be chooses with regard of light availability algorithm. There could be different approaches for sampling as initially show in simulation comparison here: #2697 (comment)

Related discussion on how light sampling works / should work #2780