fcl-db: interbase: Add support for DECFLOAT(16) data type

These patches adds base capability of loading DECFLOAT(16) fields (which are encoded in Densely Packed Decimal format). DECFLOAT(16) is mapped to FmtBCD field, but FmtBCD can not accept full range of DECFLOAT(16) values - from 1E-398 to 9.9..9E+384, so there may be an overflow exception. (so it is questionable whether not to map decfloat to double even at the cost of losing precision???) fmtbcd.pp.diff ibconnection.pp.diff So these patches can be considered as starting point ...