Extend mmCIF output to preserve tables that we do not read

[Conrad Huang]

Current mmCIF output include tables that (a) ChimeraX reads but does not change, and (b) ChimeraX uses and regenerate to reflect changes. Tables that are in the original mmCIF file but not read by ChimeraX should be included in the output file as well, since they may be generated and used by other software packages such as Phenix. Rather than storing these currently unread table separately, we can keep track of where the input file is, and reread it on output if necessary. If the input file is deleted, then the unused tables are lost.