Material theoretical density is not written to database and can be lost in restart runs

[drewj-tp]

#1440 is a large ticket that points to a broader problem that if materials have some special attributes that determine properties, those attributes are not written to database and lost in restarts. That's a large architectural problem.

A smaller, specific, and urgent problem is

A concrete example of this is the TD_frac material modifier in either UraniumOxide or B4C. Because TD_frac (and its associated instance variable theoreticalDensityFrac) is not written to the database directly, it is left at its default when the reactor is loaded back up from the database. So if you do a restart run with a case that has either of those materials in it, your material densities will be wrong if your TD_frac entries don't exactly equal the default values in the material class. For UraniumOxide that is 1.0 and for B4C that is 0.90.