Revert all arrays except wavelengths from double to single precision float

Following PIPE2D-689 it was decided to implement all arrays as double-precision floats.

Following discussions today on the slack drp-2d channel it was agreed in order to keep the data product sizes to a manageable amount, to revert all arrays except those corresponding to wavelengths to single-precision float, keeping wavelengths as double s.

Please implement this agreement.