Document data generation for unit test

[eric-czech]

See: #20 (comment)

Much like https://github.com/pystatgen/sgkit/pull/66 and https://github.com/pystatgen/sgkit/pull/76, the current pysnptools unit test uses randomly generated data from external software (Hail). This is in a validation folder within sgkit, so it may make sense to start doing the same thing in satellite repos, but it will lessen the burden for it if we wait until the multi-repo decision is final (https://github.com/pystatgen/sgkit/issues/65).

In any case, the code that generated the test data should moved somewhere central and documented. It lives here now: https://github.com/eric-czech/sgkit-dev/tree/master/io/plink