Obsolete chunks slicing

This section describes the chunk slicing strategy (support for MSET CALIBRATION BANDWIDTH) implemented in MRTCAL from its beginning up to GILDAS version apr26. It was abandoned because the overheads of 20 MHz chunks for large bandwidths where dominating the science scan calibration. The chunks now always match the hardware chunks width. The calibration bandwidth is only applied when calling ATM through TELCAL chopper_t. The calibration products (Tsys, etc) are constructed as full width chunks even if they have been computed on 20 MHz intervals. This way, science scans are completely insensitive to the calibration width: only full calibration arrays are applied, independently of the way they have been computed.