Bug that causes to load compatible but not the best possible comparison lamp

[simontorres]

Consider the following data:
Those lamps are a called a COMP Group and the the are grouped together because several keyword values are the same, among those there are radeg and decdeg which are the degree version of obsra and obsdec respectively.

         obsra        obsdec       grating  cam_targ  ...     filter filter2  gain  rdnoise  lamp_hga lamp_ne lamp_ar lamp_fe lamp_cu      radeg    decdeg
12:27:25.776  11:06:28.148  600_SYZY_OLD      20.0  ...  NO_FILTER   GG395  2.06     3.99     FALSE   FALSE   FALSE   FALSE    TRUE  186.85740  11.10782
12:27:25.776  11:06:28.148  600_SYZY_OLD      20.0  ...  NO_FILTER   GG395  2.06     3.99      TRUE    TRUE   FALSE   FALSE   FALSE  186.85740  11.10782
12:27:25.776  11:06:28.148  600_SYZY_OLD      20.0  ...  NO_FILTER   GG395  2.06     3.99      TRUE    TRUE   FALSE   FALSE   FALSE  186.85740  11.10782
12:27:25.776  11:06:28.148  600_SYZY_OLD      20.0  ...  NO_FILTER   GG395  2.06     3.99     FALSE   FALSE   FALSE   FALSE    TRUE  186.85740  11.10782
12:27:25.776  11:06:28.148  600_SYZY_OLD      20.0  ...  NO_FILTER   GG395  2.06     3.99      TRUE    TRUE   FALSE   FALSE   FALSE  186.85740  11.10782

The problem is that radeg and decdeg have a precision of two decimals and this leads to combine in the same group lamps that should be in a different one but since they are for the same target but taken at different moment they will no provide the best possible wavelength solution.

The solution is to increase the precision to 6 decimals.