Network building enhancements

[LRossman]

See ReleaseNotes2_3.md.

[eladsal]

@LRossman regarding backward comparability, is it right to say that the only thing that breaks is the assumption that if ENopen return's 0 the network data passes all the tests?

[LRossman]

@eladsal the first commit (80f9acf) that passed the CI tests contained only the updates that allowed ENopen to successfully process a partial network and to set a curve's type so that my Lazarus GUI project could move forward. These changes had no effect on the solver's results.

Then I got greedy and tried to incorporate some changes to the hydraulic solver that would improve its robustness by replacing the status checking it does during the Newton iterations with continuous penalty functions. These actually work quite nicely but as with all changes to the solver code they can produce results that are slightly different from the benchmark test results (e.g., a flow of 30.29 gpm compared with 30.30) thus causing the CI tests to fail.

Since I don't want to re-do the benchmark results at this time, I will revert dev_2.3 back to its 80f9acf state and fork another branch from it where I will continue my efforts to improve the solver.

[LRossman]

Finally getting around to merging this update into dev. It addresses issues #575, #576, and #579.