writeTableCF now removes quartet sets that were not observed in tree sample

[gaballench]

For some quartet sets we can have combinations that are not observed in the tree sample because all possible combinations are generated that may not be filled. This however represents lack of data in the sense that a given entry with 0.0 in all qs, and then also in ngenes, adds no information.

[cecileane]

It is a breaking change: what was the motivation?

• Is it obvious to users that the absence of a row for a quartet means the absence of data? versus the lack of knowledge about what the gene trees contain?
• Is it possible that the user would still want a row for any 4-taxon set that has 0 genes, so they can get summaries about how many 4-taxon sets are missing data, or how often a particular taxon appears in 4-taxon sets that lack data, etc.?

This function was moved back to PhyloNetworks and renamed to tablequartetCF, so I will close this PR.

But see this commit 83917af52 for a different implementation of your suggested change.

Still needed:

• add the option to retain these rows, and
• document the new behavior in the docstring.