Following on from #943 where a separator arg was added to from-csv.

I've now realized that from-tsv is just a specific case of from-csv (tab separated values).
I've also noticed that passing --separator "\t" to from-csv trips my uninspired check for a single character (it sees two characters, '' and 't').

I'd like to propose having from-tsv and from-csv --separator "\t" do the same thing, and I think we can deduplicate some code while we're at it. And I'm planning to make a pull request with these changes :)

1. [from/to]_tsv can call the same function as [from/tp]_csv, this has already been started with from_structured_data. Only, I think the scope of this module would be more clear if we named it from_delimited_data. This will use csv::ReaderBuilder and csv::WriterBuilder with different delimiters.

2. the parsing of the separator arg of from_csv and to_csv should be improved. I struggled to get a character out of nu from the command line and ended up checking a string.to_chars() has length 1. I can check if the string is "\t" first and return '\t' early for now but I worry that's symptomatic of a deeper issue? Are there plans for supporting characters from the command line? I'd be interested in the why/why not and how?

Also a side note: maybe --delimiter should be added as an alias for --separator?