Fix robust counting and regex for string comments

[gkarthik]

Two changes,

• Changed regex to allow parsing of robust counts when codon not present
• Regex for parsing in strings seems to be doing something unexpected.

L1069 has the following regex: [,&][A-Za-z\_\.0-9]+=["|\']*[A-Za-z\_0-9\.\+ :\/\(\)\&\-]+[\"|\']*. It is unclear what the * is doing? Removing it seems to restore expected behavior.

[evogytis]

It looks like what we had earlier with * was quite permissive and parses some comments multiple times (the correct type and then as strings for the second time). Unfortunately this pull request breaks parsing of beast2 strings (see https://github.com/blab/mers-structure/blob/master/data/figure%201-source%20data%201.zip or https://raw.githubusercontent.com/nicfel/Reassortment-Tutorial/master/precooked_runs/H1N1pdm.mcc.network.trees) which are not enclosed with quotes. I think we should start collating various trees we encounter to include in unit tests for exactly these types of scenarios.