Hlsvdpropy

------------------------------------------------------------------------------
This is a bugfix release for deprecated function calls in numpy. Also updated
URL in setup.py to describe the GitHub repository that now houses the code.
No change to algorithm functionality.

------------------------------------------------------------------------------
The default value for the SVD algorithm in hlsvdpro() has been set to use the
scipy.linalgsvd() method by setting the default for the sparse keyword to False.
The user can still choose to set it to True if they want. Reason - After using
this code on a number of data sets in the Vespa-Analysis platform, we saw
instability when data size or nsv_sought were set to 'edge' values. That is
either very small or very large. This was for data with 4096 point. So we were
able to push the values far enough to test rigorously. The fix for this seems
to be to use the 'full' scipy.linalg.svd() method and truncate the results to
the nsv_sought.

------------------------------------------------------------------------------
This is HLSVDPROPY release version 2.0.0, it is a 'pure Python' implementation
of the algorithm provided by the Fortran HLSVDPRO version 2.0.0 package.

Note. We are starting with version 2.x to be in line with the version of
HLSVDPRO to which it corresponds.

HLSVDPRO 2.x was adaptated to use PROPACK library version 2.1 to implement the
HLSVDPRO algorithm as described in:

Laudadio T, et.al. "Improved Lanczos algorithms for blackbox MRS data
quantitation", Journal of Magnetic Resonance, Volume 157, pages 292-297, 2002.

we are using the scipy.linalg.sparse.svds() algorithm to replace PROPACK.
Otherwise, the algorithm remains the same.

HLSVDPROPY version 2.0.0 is compatible with Python 2.7 and 3.x.