Shared code for the R scripts #101

Suggested by @maxulysse in #54. Several module templates copy the same R code instead of sharing it, e.g.:

• the false-doubles estimators seq_error_correct_by_false_doubles_MLE and seq_error_correct_by_false_doubles_EB share most of their code
• correct_library_completed and seq_error_correct_counts_for_heatmaps are defined in both error-correction templates (false doubles and wildtype);
• position_biases is defined in both global_pos_biases templates.

To decide together with nf-core team: shared code either in bin/ (a small R file sourced by the templates, or whole scripts), or templates that stay self-contained. The current nf-core module specification recommends templates as the most portable option, while bin/ would avoid copies.

Best done together with the planned v1.1.0 performance overhaul. In this, we will likely merge several of the current processes to reduce unnecessary scheduling.