grab should write more accurate WCS information

[StuartLittlefair]

The WCS produced by grab doesn't use the offset information which is now stored in the .fits images saved from hipercam data, which is should, so each .hcm file would have a better WCS

[StuartLittlefair]

To clarify how the offset pattern is stored in the .fits images, users define a list of absolute RA and Dec offsets, which are repeated until the end of the run. https://github.com/HiPERCAM/hcam_widgets/pull/28 updated the HiPERCAM control software so these offsets were stored in the 3rd extension of the .fits files.

grab already creates a WCS header, which is used for all .hcm files output. To fix this issue we need grab to:

• find out which nodding offset applies to this frame (these are repeated, so if there are 10 offsets, frames 10, 20, 30... would use the 10th offset
• tweak the WCS header to account for the nodding offset