Modification and optimization on qqtl and qtwas weights pipeline

[al4225]

1. Added phenotype_id column in vQTL.
2. Set default ld_clumping, ld_pruning, screen_significant of qqtl as false.
3. Once keep_variants file is provided, only calculate the qr_coef based on this file, but the qr_screening and vqtl screening will not be based on this list; still, calculate all of the variants with default screen_significant = FALSE.
4. Added XICOR function and package to perform Chatterjee correlation test.
5. Filtered p_qr < 0.05 before Step 8: Filter highly correlated SNPs in qtwas weights, otherwise, when we don't first filter variants for qtwas weights calculation, the high dimension #p will be greater than #n, which will lead to a singular matrix issue (not full rank) even after iterative corr filter until 0.5 cutoff.
6. Other optimizations