Add EN_ERRORNODE to EN_getstatistics to allow retreiving node causing ill-conditioned network

[AbelHeinsbroek]

See issue #885 for more information

[LRossman]

I have a few comments.

1. I don't see the need to introduce the hydraul.IsIllConditioned variable, since it's enough to check if hydraul.ErrNode is > 0. Plus Error 110 already indicates that hydraulics couldn't be solved.
2. Maybe I missed it, but I don't see where ErrNode is initialized hydsolver.c.
3. You've introduced a new error message, 265,"network is not ill-conditioned", which does not really indicate an error condition like all of the other error codes do. I suggest you remove it.

[AbelHeinsbroek]

Thank you for the feedback, I've made the suggested changes. I removed the new error message, and use 203 - "undefined node" instead.

[LRossman]

I don't think you want to use 203 as an error for when ErrNode = 0. ErrNode = 0 is a valid result, indicating that there is no node causing the network to be ill-conditioned. So please remove lines 1103 - 1105 in the latest commit.

Also, although you've now initialized ErrNode in hydsolver.c, I wonder if all of the variables retrieved with the EN_getstatistic function need to be initialized when a network is first created, since the function could be called before any analysis has been made in which case the variables would contain garbage. This is probably best left as a new PR.

[AbelHeinsbroek]

right, I forgot that nodes and links are not zero-indexed. I've made the changes.

[LRossman]

@AbelHeinsbroek I lost track of this PR. Does it address all of the comments and is ready to be merged? If so, I will go ahead and merge it.