Implement new variables and constraints for improved tank performance

[tasseff]

Vieira et al. propose a conditional approach that differentiates between full and non-full tanks. This results in a potentially larger range of tank behaviors and should especially be accounted for in optimal water flow problems. My recommendation is that the new formulation for tanks extend across all problem specifications.

[tasseff]

Hey @enzoliuyang94, we can discuss and document implementation details here.

[tasseff]

As discussed, we will use the tank/pipe/shutoff valve combination for the modeling of tanks, so this issue can be closed.