Add the supervised symbol detection layer

[mz2]

• [x] Build example suite of symbols for unsupervised symbol detection (single labeled example per symbol type, Dollar P score used for matching).
• [x] Automatically test existing unsupervised symbol detection with the example suite of symbols (declare success by asserting scores when comparing known stroke sequences to known templates -- this is to allow refactoring the score computation).
• [x] Build example suite of symbols for supervised symbol detection.
• [x] Finish implementing the value transformers for transforming the stroke sequences to a combo of a Dollar P score, rotation and scale invariant moments, ellipticity.
• [x] Finish implementing the random forest classifier that uses datum instances, value transformer and outputs a classification decision.
• [x] Automatically test classifying based on above features (declare success based on minimum accuracy of the classification -- strictly speaking these can fail but should be possible to either make probability low enough or to set random generator seeds to allow deterministic testing).

[mz2]

I've begun work towards this by completing the interface declarations for MPDataSet and MPDatum, and implementing concrete types (MPDataTable, MPRow).

[mz2]

I separated the term value transformer into two: kind of transformations which do not change dimensionality, and the kind described above which I've described as "dimensionality mapping". The former turned out to be much less useful than imagined at first for random forest classification than implementing dimensionality mappers which would also be used with other kinds of learning algorithms, and the second is what is used for mapping the input data into feature space used for model training.

Value transformation for instance to transform data to Z-scores or otherwise translating / scaling values are not necessary because features are independent and not made for instance linear or other combinations of internally when classifying.

[mz2]

I'd say this is done.