Generalizing quality control flags

[DSuveges]

Currently we have a neat infrastructure to flag study loci for whatever reason. However there is a need to apply dataset specific flags to other datasets in a similar manner. Eg. flagging studies that have problematic summary statistics, or lack ancestry annotation to resolve LD.

Tasks

• We need to move _update_quality_flag from study_locus to dataset.
• As study_locus inherits from datasets no other update is required.
• Refer to this function from other dataset + each dataset needs to have their own collection of QC functions and messages.

[DSuveges]

QC doesn't make sense for a set of datasets eg. overlaps. So we should move the logic to the relevent datasets instead of inherit to every one. Doing so would also imply having QC field in every schema.