Maximum # of EAM output tapes is too small for SPECIAL production runs

[tangq]

The "SPECIAL" production runs require 11 EAM history tapes as documented here. This exceeds the maximum allowed number (10 probably).

Test got errors:
ERROR: Command: '/gpfs/fs1/home/ac.qtang/E3SMv2/code/20220320/components/eam/bld/build-namelist -infile /lcrc/group/e3sm/ac.qtang/E3SMv2/test-output.v2.NARRM.amip/tests/M_2x5_ndays/case_scripts/Buildconf/eamconf/cesm_namelist -csmdata /lcrc/group/e3sm/data/inputdata -ignore_ic_year -use_case 20TR_eam_CMIP6 -inputdata /lcrc/group/e3sm/ac.qtang/E3SMv2/test-output.v2.NARRM.amip/tests/M_2x5_ndays/case_scripts/Buildconf/eam.input_data_list -ntasks 1920.0 -namelist " &atmexp /" ' failed with error 'CAM build-namelist - ERROR: Invalid namelist variable in '-infile' /lcrc/group/e3sm/ac.qtang/E3SMv2/test-output.v2.NARRM.amip/tests/M_2x5_ndays/case_scripts/Buildconf/eamconf/cesm_namelist. ERROR: in _validate_pair (package Build::NamelistDefinition): Variable name fincl11 not found in /gpfs/fs1/home/ac.qtang/E3SMv2/code/20220320/components/eam/bld/namelist_files/namelist_definition.xml' from dir '/lcrc/group/e3sm/ac.qtang/E3SMv2/test-output.v2.NARRM.amip/tests/M_2x5_ndays/case_scripts/Buildconf/eamconf'

[rljacob]

PR that increased to 10 looks pretty straightforward #3860

[beharrop]

When I made the change to 10 tapes, I stopped there because adding more would require additional effort. Specifically, in cam_history.F90 the history_readnl subroutine seems to have 10 tapes hard-coded in. At that time, we decided 10 tapes was fine to move forward with the water cycle experiments because we only had 10 tapes being requested. Looking at the Confluence page history, two new tapes were added to the special request after #3860 was put together, which is why the issue is popping up now.

[tangq]

I recall the discussion about changing to 10 tapes. It looks like we need to increase that to 15?

[tangq]

Further increasing to 15 EAM tapes is not so straightforward as #3860 as it requires modifications to the fortran code @beharrop pointed out. I took some effort to change the code as well as the namelist options. It will be fixed by #4892.