simplify configuration to 2 files

[sreichl]

• maybe reduce from 4 files to 3 or 2
• pipeline config + project config → config
• sample/unit config + metadata config (would lead to redundancy in rows, but neater) → annot
  • metadata maybe even less needed without unsupervised analysis
  • metadata is then also an output (sample x metadata)
• configurator function: check if one of the rows has skip_preprocess in the annotation file, if yes then exclude the whole sample (ie all rows of that sample) → needed still? nope, removed that column

[sreichl]

required columns:

• sample_name (first column!)
• read_type: single or paired
• bam_file: path to bam file
• pass_qc: between 0 (not used for quantification) until 1 -> maybe use this as implicit quantification flag?