Random strategy does not change it's classification

[drvinceknight]

We need to include something similar to the setup in memory one strategies that overwrite the stochastic classifier upon initialisation:

>>> p1 = axl.Random(0)
>>> p2 = axl.Random(1)
>>> p1.classifier['stochastic'], p2.classifier['stochastic']
(True, True)

Mildly connected to #779 (found it whilst working on that).

[marcharper]

Good catch