Add KING methods for genetic relationship matrix calculation

[Hua-Zhou]

For the gym function, we should add the two methods method=:KING_homo and method=:KING_robust, used in the KING software.

The methods are described in this paper: https://doi.org/10.1093/bioinformatics/btq559. :KING_homo (assuming homogeneous population) method is equations (5) and (6). :KING_robust (handling heterogenous populations) is equation (9).

[kose-y]

The current implementation of GRMs looks very memory-consuming for kinship pruning. Is there other use of GRMs other than pruning?

[juhkim111]

Maybe we can just compute and return a vector of diagonal entries and a vector of off-diagonal entries instead of the entire matrix?

[Hua-Zhou]

The current implementation of GRMs looks very memory-consuming for kinship pruning. Is there other use of GRMs other than pruning?

Can you clarify which part is memory-consuming? Calculating GRM or kinship pruning? These are two separate steps.

[kose-y]

GRM creating the whole n x n matrix as an output. Isn't a sparse matrix or a graph data structure enough as input for kinship pruning?