YMG,
I would like to revisit quadratics. It is a very conservative model for curvature. Its weakness comes by a lack of data. When applied to your own contours, you have developed a control rib with your MAX resolutions. The lack of data has been reduced.
I haven't created any quadratics that have been unfaithful to those enhanced vertices's. In actual application I would always review your MAX conditions, and probably edit them manually... I would do the same after converting them to quadratics. For myself, it's no different than deleting bogus (exterior) TIN's.
In summary, quadratics do introduce a risk that requires a critical review. Nevertheless, as a practitioner and professional, my client always expects a refined product. For them, that is visual. Smoothness. For me it is a practical accuracy. I think that your MAX contouring, combined with quadratic's, makes that happen.
It's an argument. But also only a suggestion. I think it would be nice to have Quadratic Conversion available as an automated "last step" option to production. The contours could be placed on a new layer (having a different color), and comparisons and manual editing would be easy.
In fact a greater tool for analysis would be to have a Tin-Cntr layer, and also an Avg-Cntr (for the MAX values), and a Qdr-Cntr layer (built from the MAX/AVG) which can all be reviewed to aid any manual editing that you might do to the quadratic later... or opt for using any of the others.
Rick