Add simple SGD optimiser

[neworderofjamie]

Custom update model would look something like this in standard PyGeNN (5.0.0) syntax:

gradient_descent_zero_gradient_model = create_custom_update_model(
    "gradient_descent_zero_gradient",
    params=["eta"],
    var_refs=[("gradient", "scalar"), ("variable", "scalar")],
    update_code="""
    // Descend!
    variable -= eta * gradient;
    """)

Will need to:

• Add a new class, derived off Optimiser (https://github.com/genn-team/ml_genn/blob/master/ml_genn/ml_genn/optimisers/optimiser.py). See Adam implementation for reference (https://github.com/genn-team/ml_genn/blob/master/ml_genn/ml_genn/optimisers/adam.py)
• Import into package and add to __all__ in https://github.com/genn-team/ml_genn/blob/master/ml_genn/ml_genn/optimisers/__init__.py