htsget: add samples query parameter, principally to select subset of VCF columns

[mlin]

e.g.

GET /htsget/1000genomes/variants?format=VCF&samples=NA12878,NA12877

Previously I circulated a different version of this with a repeated sample=x query parameter. This single, comma separated list is more consistent with the existing fields & tags query parameters.

[mlin]

Can a sample name have a comma in it? (@daviesrob)

[mlin]

Correct me if wrong, but it is not forbidden for a VCF sample name to contain a comma.

One possibility is to specify URI encoding of each element in the comma-separated list, so any comma within the sample name would be percent-encoded. A comma-separated list of individually URI-encoded elements seems like it would be a slightly tortured construct, though.

Another possibility is reverting to the first straw man idea of providing the list through repeated query parameters, where each individual parameter would then be query string encoded as usual, e.g.

GET /htsget/1000genomes/variants?format=VCF&samples=NA12878&samples=NA12877&samples=quick%2C%20brown%20fox

The wart is that it's dissimilar from the existing fields & tags parameters.

[nh13]

Another idea: the default delimiter be comma and then have a parameter to control the delimiter?

[mlin]

@daviesrob suggestion: percent-encoded tab delimiter

[yfarjoun]

Also, can sampleID be empty?