Refine plots

[zacharopoulou]

Main changes:

• Support experiment DE tables from either local files or Zenodo. (fixed error)
• Remove tracked Zenodo-cached DE tables from the repository intentionally.
  !!! download the updated RNA-Seq experiments from here:
• Add effect-only evaluation mode with fdr_threshold: null.
  !!! Important note here. Experiments used before where ok with their FDR value. Now, by the RNA-Seq pipeline, FDR value can be NaN or inf for some reason and in plenty occasions. Changed the logic to be able to use FDR threshold or not. If yes, then the rows without FDR are removed.
• Rename the user-facing evaluation threshold from abs_logfc_threshold to effect_threshold, while keeping backward compatibility. (created a small fix on dashboard for later)
• Improve benchmark run folder naming to include date and time.
• Optimize joined-table construction by reading only required predictor columns and filtering relevant rows during read.
• Add detailed join timing logs.
• Refine heatmap layout, gene-name y-axis labels are removed to avoid unreadable overplotting;
• Keep report threshold wording consistent for both FDR-aware and effect-only modes.

[dimostzim]

Docs still show abs_logfc_threshold in the README example, but we now expose effect_threshold

pareto plot still too busy