Add two commands

[Anaphory]

phyltr problematic: For binary trees – inspired by DensiTree complaining about some non-binary trees

phyltr sort: Takes an ordered list of taxa and tries to represent that order in the tree. In order for it to work as intended, the read_taxa probably needs to be refactored. Could it maybe be allowed to return a list instead of a set? Currently, my way to use sort with a file is with `cat file`…

[lmaurits]

I wonder if a binarise command, which replaces multifurcations with binary splits and length-zero branches, is a more elegant solution to the Densitree problem? Although I'm not opposed to a command which rejects trees with particular undesirable properties, if we can think of more of these? I initially thought "I don't like the name problematic, we should call it phyltr!", but then realised phyltr filter is asking for confusion.

Do I understand correctly, that the sort command simply attempts to swap left and right children at branch points without changing topology, to maximise similarity to a desired ordering of leaves?

[Anaphory]

I wonder if a `binarise` command, which replaces multifurcations with binary splits and length-zero branches, is a more elegant solution to the Densitree problem? Although I'm not opposed to a command which rejects trees with particular undesirable properties, if we can think of more of these? I initially thought "I don't like the name `problematic`, we should call it `phyltr`!", but then realised `phyltr filter` is asking for confusion.

Indeed. In python-newick, we had a normalizing function. That might also be useful here. But in this case, I actually needed it to check whether any of the trees given to DensiTree were non-binary (turns out they are not: The issue was that phyltr, as it currently stands, fills the slots for internal node names with the clade support, which is 1 if not specified otherwise, which meant that DensiTree thought I had multiple nodes named '1'. That is a topic for another issue.)

Do I understand correctly, that the `sort` command simply attempts to swap left and right children at branch points without changing topology, to maximise similarity to a desired ordering of leaves?

That is exactly what it is supposed to do. (If there are multiple children, it orders all of them, that's why just testing for a swap and executing it or not was insufficient)

[Anaphory]

True, tests for these commands would be useful. How does one write phyltr tests again?