Updated docs for upcoming Boise tutorial.

Tag for Boise tutorial proof read

Documentation update 1.

Updated WRF Users' Workshop tutorial documentation.

- Release 1.2.0

- Previous versions of wrf-python promoted the strings used in xarray (e.g.
  name, attributes) to Unicode strings. This caused problems when porting
  examples for PyNGL to use wrf-python in Python 3.x. All strings are now the
  native string for the Python version being used. While this should be
  transparent to most users, any users that worked with the xarray name or
  attribute values on Python 2.7 may run in to string related errors, so we've
  decided to bump the major version number.

Documentation update 2.

Fix units for pressure argument to td function.

Documentation update 1.

Release 1.1.3

Fixed/Enhanced the cloud top temperature diagnostic.
  Optical depth was not being calculated correctly when
    cloud ice mixing ratio was not available.
  Fixed an indexing bug that caused crashes on Windows, but should have been
    crashing on all platforms.
  Users can now specify if they want cloud free regions to use fill values,
    rather than the default behavior of using the surface temperature.
  Users can now specify the optical depth required to trigger the cloud
    top temperature calculation. However, the default value of 1.0 should be
    sufficient for most users.
Added 'th' alias for the theta product.
Fixed a crash issue related to updraft helicity when a dictionary is
  used as the input.
Dictionary inputs now work correctly with xy_to_ll and ll_to_xy.
The cape_2d diagnostic can now work with a single column of data, just like
  cape_3d.

Fixed OpenMP directive issue with cloud top temperature.

Release 1.1.1

- Added script for building on Cheyenne with maxed out Intel settings, which
  also required a patch for numpy.distutils.
- Fixed a few unicode characters hiding in a docstring that were causing
  problems on Cheyenne, and also building the docs with Sphinx on Python 2.x.
- Fix issue with np.amax not working with xarray on Cheyenne, causing an error
  with the mdbz product.
- Fix cape_2d private variable bug when running with multiple CPUs.

- Computational routines now support multiple cores using OpenMP. See

  :ref:`using_omp` for details on how to use this new feature.
- The CAPE routines should be noticeably faster, even in the single threaded
  case (thank you supreethms1809!).
- :meth:`wrf.getvar` now works correctly with non-gridded NetCDF variables
- The cloud fraction diagnostic has changed:
   - Users can now select their own cloud threshold levels, and can choose
     between a vertical coordinate defined as height (AGL), height (MSL), or
     pressure.
   - The default vertical coordinate type has been changed to be height (AGL).
     This ensures that clouds appear over mountainous regions. If you need
     the old behavior, set the *vert_type* argument to 'pressure'.
   - Fixed a bug involving the cloud threshold search algorithm, where if the
     surface was higher than the threshold for a cloud level, the algorithm
     would use whatever was there before (uninitialized variable bug). This
     caused some interesting visualization issues when plotted.  Now, whenever
     the surface is above a cloud level threshold, a fill value is used to
     indicate that data is unavailable for that location.
- The cartopy object for LambertConformal should now work correctly in the
  southern hemisphere.
- Fixed a bug with the PolarStereographic projection missing a geobounds
  argument (thank you hanschen!).
- Renamed the modules containing the 'get_product' routines used
  by :meth:`wrf.getvar` to avoid naming conflicts with the raw computational
  routine names. Users should be using :meth:`wrf.getvar` instead of these
  routines, but for those that imported the 'get_product' routines
  directly, you will need to modify your code.
- Fixed a uniqueness issue with the internal coordinate cache that was causing
  crashes when input data is changed to a different file in a jupyter notebook
  cell.
- Added code to better support building wheels on Windows (thank you letmaik!)
- Improved support for scipy.io.netcdf objects.
- Added a new 'zstag' diagnostic that returns the height values for the
  vertically staggered grid.
- A DOI is now available for wrf-python. Please cite wrf-python if you are
  using it for your research. (See :ref:`citation`)
- Fixed issue with vertcross and interpline not working correctly when a
  projection object is used. Users will now have to supply the lower left
  latitude and longitude corner point.
- Beginning with numpy 1.14, wrf-python can be built using the MSVC
  compiler with gfortran. WRF-Python can now be built for Python 3.5+ on
  services like AppVeyor.