Correct my middle initial.
Improve backwards-compatiblity of our use of the MxMatrix .persist slot.
Spell Hunter's first name correctly
Fix calculation of the saturated -2LL for IFA models
Add some warning about CSOLNP
Add hint about infeasible starting values (per Mike Neale)
Rework capture of errors & warnings
Permit running test from top dir
CSOLNP: Unravel some confused Matrix resizing
CSOLNP: Allow 0 coeff matrices; init to NaN instead of 1.0