[ddlm-group] Clarification on dREL tables

The dREL paper is not explicit on the uniformity of the keys used in dREL tables (associative dictionaries). In the supplementary material it says is that keys can be integers or 'character values'.  It is not explicit about whether these types of keys can be mixed in a single table. I suggest we clarify this point by stating that dREL table keys must be of the same type (the key types would be "cased strings", "caseless strings", or "integers"). I note that values are already constrained to be of the same type, and I believe that this stipulation is consistent with the current usage in the CIF dictionaries.

If there are any concerns about this, feel free to respond, otherwise I will incorporate this clarification into the Vol G dREL chapter.
--