Traverse GRG edges from disk

[dcdehaas]

In order to scale up beyond current Biobank-levels, or enable Biobank-scale analyses on non-server machines, support the traversal of a GRG that reads the edges from disk.

The initial version of this will likely only support full traversals (via node ordering) or partial traversals in the top-down direction (from mutations to samples), since the GRG only serializes down edges currently.

[yaacoo]

Another use-case would be for data data wearhouse storage and querying of genotypes by ends users, without maintaining the entire graph in memory.