Integration tests and some restructuring

[brenmous]

• Added integration tests for learn and predict commands
• Added required test data
• Output directories: outputs no longer dumped in working directory. Optional files need to have a path specified. Compulsory files that have no path specified will default to 'output' directory.
• Config object no longer pickled with model file. This was causing issues with tests and also prevents making tweaks (e.g. selecting outbands) without having to rerun the 'learn' command and retrain the model. You now need to specify a path under 'model' in the 'output' block of the config. This is where the model will be saved when running 'learn' and loaded from running 'predict'.
• Multirandomforest and multicubist 'temp' files. Training these models generates files which are required for making predictions using the model. These were previously dumped in working directory in a 'results' folder. These files aren't temporary and need to passed with the model. To make models portable these are longer saved as files but are stored in the model object itself.
• Package version is now dervied from 'git describe' command. This is also written to metadata file when making predictions.

Issues closed: close #64, close #65, close #66, close #67

[codecov-io]

Codecov Report

Merging #88 into develop will increase coverage by 27.26%. The diff coverage is 90.14%.

@@             Coverage Diff              @@
##           develop      #88       +/-   ##
============================================
+ Coverage    36.47%   63.73%   +27.26%     
============================================
  Files           32       32               
  Lines         3471     3469        -2     
============================================
+ Hits          1266     2211      +945     
+ Misses        2205     1258      -947

Impacted Files	Coverage Δ
uncoverml/scripts/gridsearch.py	0% <ø> (ø)	:arrow_up:
uncoverml/features.py	77.86% <100%> (+62.16%)	:arrow_up:
uncoverml/predict.py	33.83% <100%> (+33.83%)	:arrow_up:
uncoverml/__init__.py	100% <100%> (ø)	:arrow_up:
uncoverml/metadata_profiler.py	100% <100%> (+100%)	:arrow_up:
uncoverml/models.py	79.47% <100%> (+14.99%)	:arrow_up:
uncoverml/geoio.py	83.61% <80%> (+57.78%)	:arrow_up:
uncoverml/scripts/uncoverml.py	65.47% <81.25%> (+65.47%)	:arrow_up:
uncoverml/config.py	73.82% <88%> (+62.77%)	:arrow_up:
uncoverml/transforms/target.py	100% <0%> (+1.25%)	:arrow_up:
... and 18 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 4eba531...74a6385. Read the comment docs.