Improve documentation for risk measures, including Wasserstein

[Tangtry]

I tried the Wasserstein risk measure using the hydro-thermal example from introductory file. The codes are as follows
SDDP.train(
model,
risk_measure = SDDP.Wasserstein(Gurobi.Optimizer; alpha=0.1) do x,y
return abs(x-y)
end,
iteration_limit = 10
)
However, it fails with an output

caused by: MethodError: no method matching -(::SDDP.Noise{Float64}, ::SDDP.Noise{Float64})
The function - exists, but no method is defined for this combination of argument types.

I also tried other risk measures like Entropy and ModifiedChiSquared, these work good.

[odow]

I haven't really used this since I added it. I guess it needs better docs!

From the error message, it looks like you might need:

risk_measure = SDDP.Wasserstein(Gurobi.Optimizer; alpha=0.1) do x, y
    return abs(x.term - y.term)
end

[Tangtry]

Thanks a lot! It works smoothly with the updated code.

[odow]

I'll leave this issue open until I fix the documentation properly 😄