full/availability: Vet parameters in reconstruction tests

https://github.com/celestiaorg/celestia-node/pull/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](https://arxiv.org/pdf/1809.09044.pdf). As per @distractedm1nd, the `c` parameter does not represent the withholding case, so he made a [custom script](https://gist.github.com/distractedm1nd/d2b306da13581b97bbbcefaa5b050738) 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