Mccd

Main modifications:
- Change of the construction and handling of polynomials.
- Forcing RBF interpolation for the global part.
- Harder sparsity constraint for spatial variations.
- Dropping the normalisation proximal operator.
- Skipping the last weight optimization so that we always finish with a features/components optimization.
- Doing minor updates needed for compatibility with the new ModOpd version.
- Adding as a parameter the local polynomial degree in hybrid mode.
- Adding as a parameter the minimum polynomial degree for the global model.

Updating default config parameters:
- Increased iterations to have a good convergence in the PSF crown.
- Changed number of local/global features.
- Set denoising to zero -> the denoising is giving biased results (at least int he CFIS data).

- Fixed some minor issued with new version of the ModOpt package.
- Changed requirements.
- Minor modifications.

Changed the way the Pypi long description was handled due to a bug on setuptools that prevented the package upload to Pypi.

Now the uploaded Pypi version matches this release.

This release contains several new functionalities:

- Configuration file handling via the MCCDParamsParser class.
- Automatic runs from the configuration files via the RunMCCD class.
- Functionalities added in auxiliary_funs
- Solved minor bugs
- New example notebooks added
- New functions needed for the compatibility with Cosmostat's shape measurement pipeline.

First Release

This release corresponds to the mccd Pypi version 0.0.1.

It is working fine, still some more functionalities could be added.
The core package, the MCCD class, is included.

Up to now only one test is implemented which test the good installation of PySAP.
Travis CI working.
Automated API documentation within Travis CI.