Fix Tajima's D bug

• Fix issues running Tajima's D calculations when SciPy is not installed.

Improved Memory Management. Added tools for protein sequence alignments.

• Add ability to use protein multiple sequence alignments as input data. Useful for Shannon Entropy calculation.
• Improve memory management when calculating nearby residues. No longer creates a Euclidean distance matrix, but uses a KDTree instead.

Remove Scipy dependency and refactor calls to DSSP

• Removed SciPy dependency, which simplifies install process.
• DSSP is only called if it is required, for example when there is a need to calculate secondary structure or solvent accessibility.
• Also removed redundant methods for writing output data to file.
• Updated README to reflect optional requirements for BLAST+, Exonerate and DSSP.

Fix distribution to include CHANGES.txt

Minor fix so that the distributed code includes CHANGES.txt, LICENSE.txt etc.

Minor changes to documentation and distribution

This release adds minor changes to documentation and distribution.

Version 0.2.0 release

Added features:

• Additional genetic tests (Watterson's theta, nucleotide diversity)
• Added multi-chain support. This allows mapping of data from different chains.

Some code refactoring was performed. map method is now available from the Structure class, not each individual Chain.

Data requirements altered slightly. Each data object should be associated with a particular set of structure chains e.g. {('A', 'B'): data_1, ('C',): data_2}

Reference sequences should be provided for each chain: {'A': seq_1, 'B': seq_1, 'C': seq_2}