Pyfaidx

- Small internal refactoring

- Faidx and Fasta strict_bounds bounds checking logic is more correct
- Fasta default-seq parameter now works
- CLI script faidx now takes a BED file for fetching regions from a
fasta

- Faidx no longer has raw_index attribute or rebuild_index method
(reduce memory footprint)
- Faidx index memory usage decreased by 31-40%
- .fai creation is streaming, performance increase for very large
indices
- Possible speed regression when performing many small queries using
Fasta class

- Fasta and Faidx can take default-seq in addition to as_raw,
key_function, and strict_bounds parameters.
- Fixed issue [20](https://github.com/mdshw5/pyfaidx/issues/20)
- Faidx has attribute raw_index which is a list representing the fai
file.
- Faidx has rebuild_index and write_fai functions for building and
writing raw_index to file.
- Extra test cases, and test cases against Biopython SeqIO

- Faidx index order is stable and non-random

- Fixed a bug affecting Python 2.6