checking missing values in LD matrix:
- will cause an error with 'var_r' is not a logic value

remove the construction of the diag LD matrix to save the memory.

fix error for get_cos_summary

slient the warning message from qvalue function, and if warning happens, try another qvalue estimation of pi0

The issue: if we have multiple sumstat and only one superset LD matrix, the index of focal outcome will be mismatched.

1. dict is designed by X and LD, but we need additional dictionary for keep variables.
2. keep variables for individual level data is based on X - there is no issue.
3. keep variables for sumstat data is based on sumstat (not LD) - there is a issue to index the focal outcome based on dict later.

Keeping and commenting out the original code for initial testing, later we will remove the commented code if pass the test

Additionally:
1. adding unit test for the above issue.
2. debuging the data loader error.

Handle error since the best joint update jk_star is not included in the specific dataset. We will not consider LD for this update.

Adding uCoS filtering cirtieria
- check_null_max_ucos

adding horizontal and vertical spaces of text labels.

fix error when variable names in different datasets have different orders (e.g., dataset1 has A then B, but dataset2 has B then A).