Running MERRA2 GOCART past 2017-01-25 is impossible?

[mathomp4]

I'm not sure of the "good" way to solve this, but it seems like running GEOSgcm with MERRA2 GOCART past 2017-01-25 might be impossible. Looking at the ExtData.rc files we have things like:

GEOSchem_GridComp/GOCART_GridComp/BC_GridComp/MERRA2/20000401-present/BC_GridComp_ExtData.rc

Line 7 in 4c0f2c3

BC_BIOMASS NA N Y %y4-%m2-%d2t12:00:00 none none biomass ExtData/PIESA/sfc/QFED/v2.4r6/Y%y4/M%m2/qfed2.emis_bc.005.%y4%m2%d2.nc4

Why? Well, v2.4r6 of QFED only goes up to 2017-01-25:

(1126) $ ls -ltr /gpfsm/dnb52/projects/p10/gmao_ops/fvInput/g5chem/sfc/QFED/v2.4r6/Y2017/M01 | tail -6 
-rw-r--r-- 1 dao_ops g0620 265915 Feb  1  2017 qfed2.emis_so2.005.20170120.nc4
-rw-r--r-- 1 dao_ops g0620 275617 Feb  1  2017 qfed2.emis_so2.005.20170121.nc4
-rw-r--r-- 1 dao_ops g0620 272417 Feb  1  2017 qfed2.emis_so2.005.20170122.nc4
-rw-r--r-- 1 dao_ops g0620 278369 Feb  1  2017 qfed2.emis_so2.005.20170123.nc4
-rw-r--r-- 1 dao_ops g0620 273991 Feb  1  2017 qfed2.emis_so2.005.20170124.nc4
-rw-r--r-- 1 dao_ops g0620 276476 Feb  1  2017 qfed2.emis_so2.005.20170125.nc4

@bena-nasa believes this caused an issue for @sdrabenh in a run recently. And if I try to run in 2019:

(1091) $ rg pe=00001 1day.different-submod.out -M0
1659:pe=00001 FAIL at line=02225    MAPL_ExtDataGridCompMod.F90              <From ExtData/PIESA/sfc/QFED/v2.4r6/Y%y4/M%m2/qfed2.emis_bc.005.%y4%m2%d2.nc4 could not find time no extrapolation>
1660:pe=00001 FAIL at line=01046    MAPL_ExtDataGridCompMod.F90              <status=1>
1661:pe=00001 FAIL at line=01839    MAPL_Generic.F90                         <needs informative message>
1662:pe=00001 FAIL at line=00725    MAPL_CapGridComp.F90                     <status=1>
1663:pe=00001 FAIL at line=00566    MAPL_CapGridComp.F90                     <status=1>
1664:pe=00001 FAIL at line=00870    MAPL_CapGridComp.F90                     <status=1>
1665:pe=00001 FAIL at line=00496    MAPL_Cap.F90                             <status=1>
1666:pe=00001 FAIL at line=00364    MAPL_Cap.F90                             <status=1>
1667:pe=00001 FAIL at line=00205    MAPL_Cap.F90                             <status=1>
1668:pe=00001 FAIL at line=00182    MAPL_Cap.F90                             <status=1>
1669:pe=00001 FAIL at line=00032    GEOSgcm.F90                              <status=1>

Now the solution is "obvious" to me: Use QFED v2.5r1 which seems to be up to modern day. But, that means (I'm guessing) pre-2017 runs of GEOSgcm with MERRA2 emissions are instantly non-zero-diff.

The only other thought I can think of is to make a "meta" QFED symlink directory a la MERRA2_MEANS where if we know that real MERRA2 switched from v2.4r6 to v2.5r1 on day YYYY-MM-DD then a directory tree of symlinks could move from v2.4r6 to v2.5r1 on that day to maintain symmetry?

Of course, this is waaaaaaaaaay above my pay grade. I'm guess @adarmenov is the best person to ask about this, so I've tagged him. :)