Hotfix: fixes #140

[bena-nasa]

Fixes #140
Also needs this PR from chem as well to fix the issue:
GEOS-ESM/GEOSchem_GridComp#182

[adarmenov]

@bena-nasa the way this PR and the accompanying ACHEM PR were formulated, e.g. 'Variables being provided by ExtData must be MAPL_RestartSkip' makes me wonder if the aforementioned requirement breaks the symmetry between variables provided by ExtData or 'model' components. Could you please clarify, if we need to differentiate between ExtData imports and non-ExtData imports?

@adarmenov I'm not sure what you are asking, these changes I've made here and in ACHEM are inline with how we have handled all the other variables that have been provided via ExtData every since we started using ExtData.

There's no way for the component to know if an IMPORT it request will come from ExtData or be satisfied by a connectivity higher up in the MAPL hierarchy, that's the point of the componentization. It's up to the user to make sure they have the right Restart option to the MAPL_AddImportSpec call. We were handling this right in GOCART1g, GEOSChem, TR, GMI etc... This just got overlooked when GOCART2G was created and since I guess no one really ran achem with replay before Achem got turned on by default it also got missed there.

[amdasilva]

It looks resonable and appropriate as it implements the same behavior in Legacy GOCART