Suggested additions to Polymer for Sen2Water

[martin-boettcher]

Dear François,

this pull request contains a few updates I did to integrate Polymer into Sen2Water. Please, feel free whether you want to take some of them on board or not. I can keep changes required for Sen2Water in a branch of our fork.

The changes are:

• level1_netcdf.py with support for Sentinel-2 resampled to 60m with a different metadata structure and top-level variables for ancillary data
• copernicus_dem.py with an implementation for dynamic download and nearest neighbour interpolation of the Copernicus 90m DEM and Copernicus 30m DEM.
• Support for reading the cloud mask from NetCDF instead of HDF
• two small fixes (multiprocessing failed at least under Linux, handling of non-finite inputs

Best regards,
Martin

[fsteinmetz]

Thanks @martin-boettcher !
I have pushed all these commits, except the one related to LUT interpolation in case on non finite inputs. Generally there is an index search through object Idx before the interpolation, which should raise a ValueError in case of inifinite values. Errors can be avoided by passing fill_value='extrema' to the appropriate Idx, or filtering out entirely infinite values from the start. In which context did you encounter this issue ?

[fsteinmetz]

I have included d7b9f08 and 21fce85 in polymer master branch.