DecisionTreeFactor prune method

[varunagrawal]

This PR adds a new prune method to DecisionTreeFactor which will allow for managing exponential growth.

Given a DecisionTreeFactor

DecisionTreeFactor:
 f[ (1,2), (2,2), (3,2), ]
 Choice(3)
 0 Choice(2)
 0 0 Choice(1)
 0 0 0 Leaf    1
 0 0 1 Leaf    2
 0 1 Choice(1)
 0 1 0 Leaf    3
 0 1 1 Leaf    4
 1 Choice(2)
 1 0 Choice(1)
 1 0 0 Leaf    5
 1 0 1 Leaf    6
 1 1 Choice(1)
 1 1 0 Leaf    7
 1 1 1 Leaf    8

if we set maxNrLeaves to 5, which means only keep the top 5 leaves, we get

DecisionTreeFactor:
 f[ (1,2), (2,2), (3,2), ]
 Choice(3)
 0 Choice(2)
 0 0 Leaf    0
 0 1 Choice(1)
 0 1 0 Leaf    0
 0 1 1 Leaf    4
 1 Choice(2)
 1 0 Choice(1)
 1 0 0 Leaf    5
 1 0 1 Leaf    6
 1 1 Choice(1)
 1 1 0 Leaf    7
 1 1 1 Leaf    8

as expected.

[dellaert]

cool

[varunagrawal]

Done and done