File format V4: shrink strings, add variant ids (backwards compat)

[dcdehaas]

This is a backwards compatible file-format change. I.e., the IGD parser will load both V3 and V4 files. V4 changes:

• The lengths of strings is encoded as 32-bit integers instead of 64-bit. We don't expect any strings longer than 4 billion chars..
• Similar to individual IDs, variant IDs can be copied over from VCF files. Generally I'd recommend putting all meta-data in a separate file (JSON or similar). You can do everything you need by selecting a set of variants/samples to operate on, then use the IGD data, and then post-process/display results using the separate metadata. There is no reason (other than convenience of having a single file) to keep meta-data with the actual genotype data.