Pyplink

Minor improvement of performance.

The only change in this version is that `PyPlink` does not load the BED file in memory. It parses the file to iterate over all markers, or seeks to a specific marker and reads its genotypes.

This help when working with huge files such as the 1000 Genomes Phase 3 dataset.

Here are the changes in this new release of PyPlink.
- Possible to write binary pedfiles (BED file) in both _SNP-major_ and _INDIVIDUAL-major_ format. For now, it is not possible to read the INDIVIDUAL-major files.
- Better testing using _Plink_.

Here are the changes in this new release of `PyPlink`.
- Possible to write binary pedfiles (`BED` file).
- Father and mother identification numbers are now always strings.

This is a minor release of `pyplink`. Here are the changes:
- The `fid` and `iid` columns in the _fam_ `DataFrame` are now always strings.
- The `snp` column in the _bim_ `DataFrame` are now always strings.

This modification was necessary, as one would expect those columns to be strings, even though they might include only integers. When looking at sample (or marker) identification numbers, one rarely cast these values to `int`...

This is a minor release of `pyplink`. Here are the changes:
- Now uses the `six` module for more efficient use of `range` in both _Python2_ and _Python3_.