only 1 run/scenario for some models

MRI has 5 runs for ssp245 but 1 for the others. GFDL has 3 runs for ssp245 but 1 for the others. MPI has 10 runs for ssp270 but 2 for the others.
This isn't ideal, and there is something we could do about it:

- [ ] we have the data for the low resolution MPI model, which has 10 runs of each scenario. i had originally preferred the high resolution version, but have since changed my mind and think that more runs trumps higher resolution. we could add MPI-*-LR to the database and replace MPI-*-HR with MPI-*-LR in the CCISS ensemble.

- [ ] i checked the [ESGF CMIP6 download page](https://aims2.llnl.gov/search) and MRI now has 5 runs for all scenarios. we could download these additional runs and add them to climr.

- GFDL doesn't have additional runs on ESGF. nothing to do here, but i think we can tolerate having one model in the ensemble with only one run in ssp126 and ssp370.

This isn't a critical need but we should probably make these changes in the next month or so.