Changes to Dataset class

[BrianWilson1]

Rename Dataset class to SciDataset.
Add method so sciDataset object can write itself to a netCDF4 file.

[rahulpalamuttam]

@BrianWilson1

Writing to netCDF4 is going to be a challenge because the user needs to install the netCDF4 C library to their systems. The java api to write netCDF4 is dependent on this requirement.

See here :
http://www.unidata.ucar.edu/software/thredds/current/netcdf-java/tutorial/NetcdfWriting.html

We can write to netCDF3 just fine, and is what the above pull request does.

[kwhitehall]

@rahulpalamuttam how about making the installation of the NetCDF-4 C Library a dependency of SciSpark (to be noted in the README.md file to install), and provide the support?

[rahulpalamuttam]

@kwhitehall @BrianWilson1 @pzimdars
We can do that, but that library needs to be installed on every node in the cluster.
If we ever get to point where we're running on a cluster with a large number of nodes - it's going to get very cumbersome.

This brings up a more important point.
There should be some work towards a SciSpark install script.
Here's https://www.ansible.com/ which we could use.

With respect to the PR - I think users would much rather prefer having to just build SciSpark and run, without installing C libraries manually. What I can do is having a switch to toggle between writing netcdf3 and netcdf4. For the netcdf4 case users must have the C library installed.

This would require updating travis ci to have additional installation procedures so our builds pass.

[rahulpalamuttam]

@kwhitehall @BrianWilson1

I'm having a tough time getting Java to see the netcdf4 library installed.
I'm following the instructions here : http://www.unidata.ucar.edu/software/thredds/current/netcdf-java/tutorial/NetcdfWriting.html

Are there any members of our team who have done this successfully and used the java api to write netcdf4?