tskit-dev
First draft of a simplification tute.
Not yet finished, but will close #52 when it is. Just making this PR now so that people can look and comment.
Who putting together my ARG poster (https://docs.google.com/presentation/d/1_IcShje2CFbEPbUzodNkgNBv2m37xGx32np_YK7LvfE), I realised that there's a difference between the sort of "simplification" that keeps all nodes present but simply changes edge intervals (i.e. converting a "direct inheritance" graph into a "sample resolved inheritance" graph, and the sort of simplification that we normally mean in tskit, in which non-ancestral nodes are dropped (and potentially, ancestral unary nodes of various sorts are dropped too.
I wonder if we should have a separate word for the first thing, with the assumption that the first stage of running "true" simplification is to ensure that the edges are sample resolved, so that you can identify which nodes can be dropped, etc?
Sorry not to have merged this yet @gtsambos - I thought maybe it was worth waiting until @GertjanBisschop had finished thinking about the logical arrangement of the simplify flags.