Add RCP45 and some metadata to CESM1-LE catalog

[mnlevy1981]

Catalog now contains ctrl_branch_year, ctrl_experiment, and ctrl_member_id as
well as pointing to RCP4.5 runs in
/glade/collections/cdg/data/cesmME/CESM-CAM5-BGC-ME

[mnlevy1981]

Fixes #48

@andersy005 -- if you have notebooks that relied on LENS data already, I'd appreciate it if you could re-run them and make sure everything still looks okay. This PR should add one more experiment (RCP45) and change some of the dataframe columns but still point to all the same files.

@matt-long -- last week we talked about the YAML file, and how it would be nice to trace experiment members all the way back to the control run (perhaps by letting ctrl_* be a list?), but I think we agreed to add that feature in a different PR. I wouldn't mind an extra set of eyes on the YAML, though, to make sure the metadata we do provide is correct.

[matt-long]

I think it is misleading to call this ctrl_member_id since the run it refers is not a control run.

CMIP has conventional meta data like this

:parent_activity_id = "CMIP" ;
:parent_experiment_id = "esm-piControl" ;
:parent_mip_era = "CMIP6" ;
:parent_source_id = "CESM2" ;
:parent_time_units = "days since 0001-01-01 00:00:00" ;
:parent_variant_label = "r1i1p1f1" ;
:branch_method = "standard" ;
:branch_time_in_child = 674885. ;
:branch_time_in_parent = 18250. ;

One way to think about this is to ensure that a search comprised of the parent_* attributes returns the right thing. So we could have

parent_experiment_id
parent_member_id

and

branch_time_in_child
branch_time_in_parent

We could consider

branch_method

too and would want to describe a controlled vocabulary:
"standard"
"pertlim"
....
?

[mnlevy1981]

I definitely like parent_* and the branch_time_in_* variables, in addition to adding them here I'll try to find time to go back and update the CMIP catalogs used in cesm2-marbl as well.

Instead of branch_method, maybe it makes sense to also include a pertlim field that is 0 by default? So the 20C LENS experiments will have pertlim defined in the yaml file while the the RCP experiments won't? Then the data table will show pertlim=0 for the RCP experiments, which is an accurate statement and seems as good a way as any to indicate the members differ in where they branched from rather than by perturbations in the initial conditions. I'm open to other suggestions, but don't see any major flaws with that plan.

[mnlevy1981]

Rather than continue here, resolution to #64 will address this