High memory usage when calculating chunked Hamilton-Kerr relationship matrix

[timothymillar]

This is a bug in pedigree_kinship when used with method='Hamilton-Kerr', return_relationship=True and chunks=.... Converting the (chucked) kinships to relationships requires multiplying by a combination of the (potentially non-equal) ploidies. This is currently being achieved with a non-chunked matrix resulting in high memory usage when evaluating a single chunk of the returned relationship matrix. The offending code block is here:
https://github.com/sgkit-dev/sgkit/blob/main/sgkit/stats/pedigree.py#L1202

This should be simple to fix with a chunked ploidy matrix.