Improve modeling of tank dynamics to be better

[samhatchett]

[LRossman]

I was partially successful in adapting the methods of Todini (2011) and Avasani et al. (2012) for producing more stable tank dynamics with EPANET's hydraulic solver. I was able to reproduce the solutions obtained for the test networks in both papers. However, for other networks when a tank became full the method became stuck and would not allow the tank to start emptying again. (The above papers allow a full tank to spill any excess inflow which is not consistent with EPANET's convention of shutting off inflow to the tank.) Therefore this issue remains unresolved.