error during SNaQ score calculation: "strange qnet", "Bad Diamond I" and "qnet should have only one of the gammaz"

[cecileane]

This is from a network obtained with SNaQ (thanks to Damien Esquerré). The error comes when trying to re-calculate the expected quartet CFs for a goodness-of-fit test. Here is a reproducible example, using the attached file tableCFs.txt:

qCF = DataFrame!(CSV.File("tableCFs.txt"));
netstring = "((a,b):0.741,((((((c,(d)#H29:::0.893):0.039,(e,#H29:::0.107):0.039):1.682,((f,(g,#H25:0.0::0.331):0.519):0.591,((h,i):1.058,j):0.128):0.185):0.105,k):0.126,(((((l,(m)#H1:::0.897):0.714,(#H1:::0.103,((n,o):0.502,p):0.076):0.165):1.97,((q,((r,((s,t):0.499,(u)#H28:::0.765):0.199):0.134,(v,#H28:::0.235):0.437):1.889):0.0)#H30:0.308::0.976):0.247,#H30:0.0::0.024):0.367)#H25:0.003::0.669):1.142,(w,#H27:::0.273):0.25):0.24,(x)#H27:::0.727);"
net = readTopology(netstring)
topologyQPseudolik!(net,readTableCF(qCF)) # recalculate SNaQ score and all expected CFs

and here is the output with error:

between 49.0 and 838.0 gene trees per 4-taxon set
ERROR: strange qnet when net has node 5 Bad Diamond I: qnet should have only one of the gammaz if it does not have node, but it has two
Stacktrace:
 [1] error(::String) at ./error.jl:33
 [2] updateHasEdge!(::PhyloNetworks.QuartetNetwork, ::HybridNetwork) at /Users/.../src/pseudolik.jl:445
 [3] extractQuartet!(::HybridNetwork, ::Quartet) at /Users/.../src/pseudolik.jl:492
  ...

The error is strange, in that it goes away easily. I have a hard time providing a simpler example. Here is a function to remove some taxon set from both the network and the data, to see if the error persists.

function rmtax(taxa; unroot=true)
  qCF = DataFrame!(CSV.File("tableCFs.txt"));
  net = readTopology(netstring)
  for taxon in taxa deleteleaf!(net, taxon; unroot=unroot); end
  filter!(row -> !any(row[i] in taxa for i in [:t1,:t2,:t3,:t4]), qCF)
  return topologyQPseudolik!(net, readTableCF(qCF))
end
rmtax([]) # same error as above: original full taxon set
rmtax(["x"]; unroot=false) # error after removing taxon x, without unrooting the network
rmtax(["x"]) # no error!!
rmtax(["h"]) # error
rmtax(["h","i"]) # no error