Peptides

Added
- `Peptide.svger_descriptors` to compute SVGER descriptors from Tong *et al.* (2016).

Fixed
- `Peptide.physical_descriptors` not being computed by `Peptide.descriptors`.

Documentation
- Refactor documentation of individual descriptors into their own class.
- Add missing `PCPDescriptors` and `PhysicalDescriptors` to the API documentation.

Fixed
- `peptides.datasets` data files missing from the source distribution.

Added
- `Peptide.linker_preference_profile` to build a profile like used in the DomCut method from Suyama & Ohara (2002).
- `Peptide.profile` to build a generic per-residue profile from a data table ([3](https://github.com/althonos/peptides.py/issues/3)).

Fixed
- `Peptide.hydrophobic_moment` not working properly on sequences smaller than the provided window length ([1](https://github.com/althonos/peptides.py/issues/1)).

Added
- `Peptide.counts` method to get the number of occurences of each amino acid in the peptide.
- `Peptide.frequencies` to get the frequencies of each amino acid in the peptide.
- `Peptide.pcp_descriptors` to compute the PCP descriptors from Mathura & Braun (2001).
- `Peptide.sneath_vectors` to compute the descriptors from Sneath (1966).
- Hydrophilicity descriptors from Barley (2018).
- `Peptide.structural_class` to predict the structural class of a protein using one of three reference datasets and one of four distance metrics.

Changed
- `Peptide.aliphatic_index` now supports unknown Leu/Ile residue (code *J*).
- Swap order of `Peptide.hydrophobic_moment` arguments for consistency with profile methods.
- Some `Peptide` functions now support vectorized code using `numpy` if available.

Initial release.