optimization on qr_and_twas.ipynb

[al4225]

1. Added qrank_screen_calculate parameter to skip qr_screen
2. modified variant_list strategy,

• if there is only one column (variant_ids), then only calculate qr marginal coefficient based on this list;
• If multiple columns are provided with colnames as context molecular_trait_object_id variant_id, then filter based on context-molecular_trait_object_id group, and extract unique variant_id for that group. If there is no matched pair, then skip the marginal coefficient calculation step.
• This option allows us to only provide one significant file or separate files into multiple variant_ids files for diff context-molecular_trait_object_id group.