Investigate bounding errors on fixed variable

[mo-laurenboon]

During production run of suite u-ea789 the following error occurred or fixed variable:

Unable to produce MIP requested variable "sftlaf_ti-u-hxy-u" for "CMIP7_land": No data available for "2100"; please check run_bounds
No data available for "2100"; please check run_bounds

This should be investigated and resolved where possible.

[mo-laurenboon]

Note that this variable was previously produced for UKCM2 without issues, the ancil file qrparm.landfrac.pp which I'm assuming is the one used here has a single line difference between UKCM2 and UKESM1-3 for stash code 505. This variable also uses stash 216 (I am assuming from slthick_rootd_sftgif.pp) but this setup appears consistent between models.

UKCM2-0-LL:

       START          END                                                      
   Y M D h m    Y M D h m   PROC STASH   EXP     LREC LBLEV       LEV     HLEV 
000000000000 000000000000      0   505 AAAAA    27648  9999      0.00      0.0 
Read     1 field.

UKESM1-3-LL:

      START          END                                                      
   Y M D h m    Y M D h m   PROC STASH   EXP     LREC LBLEV       LEV     HLEV 
000000000000 000000000000      0   505 AAAAA    27648     0      0.00      0.0 
Read     1 field.

[mo-laurenboon]

Note this has also occurred in u-eb409

[mo-gill]

Seems that this problem was caused by lbtim=1 in slthick_rootd_sftgif.pp . I remember seeing a during ancil testing that cdds/iris doesn't handle ancils where there's a time dimension indicated on them.

Have edited the ancil so lbtim=0 (time invariant/constant) and tested it. It's now producing for UKESM1_3.

Loading data for "CMIP7_land.json: sftlaf_ti-u-hxy-u"
Successfully produced "CMIP7_land.json: sftlaf_ti-u-hxy-u"

The ancil has been moved into the ancil dir for that model, and the broken one moved to the old folder. I've also added a README in there to keep track of deprecated ancils with the above explanation.