Added man page for mxThreshold and one factor demo
As discussed at last week's meeting, added a test to models/passing to check that both NPSOL and CSOLNP throw Code Red when indeed they both should.
Do not resetDataSortingFlags
I hope this change will not break backward compatibility. If the data is
sorted, why do we need to mark it unsorted after mxRun? Sorting it again
should obtain the same result as not sorting it again.
Remove deadcode from failed experiment
CSOLNP: taking care of general warnings
CSOLNP: warnings about convergence solved
Can omit initial fit computation after the first iteration
Add some args for control of NA behavior
Random number generation changed from R-2.14 vs later versions