Add some more simple gradient descent methods

[MichaelClerx]

Part of #684
E.g.

• gradient descent with momentum Adding Adam optimiser #1460
• adaptive learning rate gradient descent, e.g. AdaGrad AdaGrad optimiser #1468

(Note both the links are for stochastic gradient descent, but we'd be implementing a simpler, non-stochastic one)

Won't be brilliant, but can be very informative, so largely an education thing this (or just fun)

[MichaelClerx]

We have Adam & iRprop now, which meet the bill.
Also tested AdaGrad (#1468) and AdaDelta (#1470) but they didn't look promising