test_json_rmf.txt

A FILE HAS A TYPE (PHA, RMF ETC) - ALLOWED ONE OF THE DEFINED TYPES
    A FILE TYPE CONTAINS ONE OR MORE DEFINED EXTNAMES
        A DEFINED EXTNAME HAS
            ALTERNATE FORMS ("SPECRESP MATRIX" OR "MATRIX")
            A FITS TYPE (BINTABLE, IMAGE)
                ONE OR MORE KEYWORDS
                    WHICH HAVE REQUIREMENT TYPES (REQUIRED/RECOMMENDED), DEPENDENCIES (MJDREFI REQUIRES MJDREFF etc),
                    DATATYPES (STRING, INTEGER, ETC), COMMENTS
                NONE OR MORE COLUMNS; COLUMNS HAVE:
                    TYPES (REQUIRED, RECOMMENDED), DEPENDENCIES, DATATYPES, KEYWORD FORMS,
                    COMMENTS, UNITS, DISPLAY FORMATS, NULL VALUES


So for example:

RMF:[SPECRESP:{
                KEYWORDS
                COLUMNS
                }
     EBOUNDS:{
                KEYWORDS
                COLUMNS
                }
     ]

So suppose we have an HDU object in pyfits that we've read in from some file;
    we then create a template hdu for the FILE TYPE of interest (PHA, RMF, ETC)
    then for each header keyword in the template HDU, we can test whether:
        a) the header keyword exists in the test file
        b) the header keyword has the correct datatype
        c) the header keyword has a useful comment
        d) any dependent HEADER keywords exist

        then we could do the same for the columns (but we'd need to check if the keyword
        version of the column exists if the column is missing)