add realized user demand and expose via BMI

[SouthEndMusic]

Fixes #1098.

@visr @HendrikKok would it be OK if realized is not an average but a cumulative volume and resetting it to 0.0 is the responsibility of the coupler? Otherwise we need a discussion about where in the Ribasim code these things must happen in such a way that dividing by the time interval happens before realized is accessed and resetting afterwards.

Also good to know that demand is per user per priority, but realized is only per user.

For BMI variable name upgrade instructions see:

https://github.com/Deltares/Ribasim/pull/1190/files#diff-88d699c34d26c16edb268bdacbda2afb8dad704b8db2931a83e04b9ca52dc31a.

[visr]

would it be OK if realized is not an average but a cumulative volume and resetting it to 0.0 is the responsibility of the coupler?

I think that would be ok. But realized should (in a separate PR) also be added to the results. That can only work if we save two copies of the vector in memories, one for BMI and one for results, to prevent setting it to zero every saveat. That is why we have #1189. With that one vector is enough, and we can directly serve averages over BMI.