Create error, warning, or ignore system for nml params that conflict

[jordansread]

It seems failing early on setting params incorrectly would be helpful for users, especially with GLM errors aren't the easiest to track down. Starting with really simple ones first to create a pattern for implementing these rules would be ideal, making it easier for others to propose and implement more complex ones in the future.

For the time being, I am thinking about a couple of things:

• Conflicts between lake_depth and the_depths where we should fail if max(the_depths) > lake_depth
• variables that specify the length of a vector, but that vector is a different length (in this case, GLM will ignore the values after the specified length), such as bsn_vals and its relationship to H and A. (Others are csv_point_nvars, num_depths, num_inflows, inflow_varnum, num_outlet, n_zones and probably additional...). If using a default, perhaps these should warn instead of error?

Seems like a good starting point. Would be nice to be able to avoid errors or warnings if you want, so that should be coded in the validation checker.

[jordansread]

Another tricky one that will cause a GLM v3.1 segfault but not let you know why: when crest_elev is below the bottom of H