Assert that input chracters are in the alphabet when reading in a TCM file

[recursion-ninja]

When we read in data files and associate them with a TCM, they inherit the alphabet defined in the TCM file.

We need to assert that every symbol found in the input files are symbols defined in the alphabet of the TCM file.

If we don't do this at "unification time," we can get very, very cryptic errors with empty dynamic character elements when decorating the tree.