Make NaiveMeanEffects fall back to the NaiveTissueDrugMeanPredictor for LCO #343
Description
After @nictru added the new naive tissue drug mean predictor (#334), we should update the NaiveMeanEffects predictor to fall back to it in the LCO mode/when no cell line information is available. This would make the normalized metrics more faithful, as remembering tissue effects would be excluded from the correlations.