Fix issue#69
Introduce unittests for HaSa.
Avoid global variables.
Introduce a new way to rerun aborted classification.
The installation is now more light since all the packages for MiSAC are optional.
Update documentation.

Fix issue #64

Remove pinned version for BH
Predictors name should be the same in the raster and reference data.

Fix documentation for HaSa 2.0.0

Fix some links in the documentation. 
Improve progress bar.
Fix a variable initialization

Optimized version of HaSa
1. We are now able to define the number of trees for randomForest.
2. Run in optimized mode 
  a. Improve the code to only use raster objects in memory. Clean the temporary storage.
  b. Use matrices instead of raster when possible.
3. Add a new sample method which only works for the optimized_mode since it runs with matrices.
4. Add `last_ref_val`, this is, the default reference value for the last step (default: 500)

Please check the documentation.

Fix bug in the plot_results function.

Bumped version to v1.3.4 - fix bug in the save_samplepoints function.

Fix bug #50 and partially issue #44 (do the prediction using the raster that is in memory).

Bumped version to v1.3.2 - Fix bug in plotting results.

Bumped version v1.3.1

Improve how the kml files and sample points are saved into disk.
Fix a bug in checking if the results directory is empty or not.