Optional: Internal reference cohort

[skanwal]

Introduce an option to use or not use internal reference cohort. Check if --batch_rm will be effective for this.
For us it consists of 40 internal pancreatic samples that might not be useful for other cancer types.

[JMarzec]

The --batch_rm parameter set to TRUE triggers the batch_effect_correction chunk (https://github.com/umccr/RNAsum/blob/main/inst/rmd/rnasum.Rmd#L903C7-L903C30) BUT it doesn't prevents from using the internal reference cohort for data transformation, filtering and normalisation (the patient counts data is combined with the internal reference cohort for these, see https://github.com/umccr/RNAsum/blob/main/vignettes/img/counts_post-processing_scheme.png and https://github.com/umccr/RNAsum/blob/main/R/refdata.R#L24). I think that the --batch_rm parameter:

1. should be set to FALSE as default (given limited benefit of performing batch effects correction we decided to skip that step)
2. should prevent from using the internal reference cohort in general. Instead, the sample counts data should be combined with the external reference cohort (TCGA) and then collectively transformed, filtered and normalised.

[JMarzec]

Given that the batch effects correction step will be set to FALSE we also need to update the following:

1. README, in particular the "Internal reference cohort" section
2. RNAsum data processing workflow, including the Figure (remove the "Internal" reference cohort within step 2).
3. counts_post-processing_scheme.png (use the updated version attached here)
4. Z-score_transformation_gene_wise.png (use the updated version attached here)
5. Z-score_transformation_group_wise.png (updated figure needs to be generated)
6. centering_group_wise.png (updated figure needs to be generated)