Support a naming convention for the `INSTANCENAME` that can be set differently by different organizations deploying ASGS

[jasonfleming]

We've found that different organizations that use ASGS have different needs and preferences for the ASGS INSTANCENAME.

Originally, this parameter was set on an ad hoc basis by different Operators to prevent collisions between their own ASGS instances. As a result, asgsh and external tooling in/around it assumes the INSTANCENAME is hardcoded in the ASGS config file.

Experience has shown that Operators would like to have this parameter set automatically according to different ASGS instance characteristics that may include the mesh, forcing, machine name, operator ID, team ID, project ID, etc.

ASGS should support an easily specified default naming convention for ASGS instances that can optionally be stored in local assets by different organizations.

Related to #1602 #1598 #1502 #1527

[wwlwpd]

Here I linked them all with a label, https://github.com/StormSurgeLive/asgs/issues?q=is%3Aissue%20state%3Aopen%20INSTANCENAME

[wwlwpd]

I think we have a few options.

One of them is to define a global function (e.g., _get_INSTANCENAME) that has available fields that can be defined in the environment as INSTANCENAME_FMT that is a sprintf type way to define the string (e.g., %M = machine name, %E is ASGSADMIN email, or %Iis ASGSADMINID, %G is gridname, etc).

We could define a default INSTANCENAME_FMT, but this could redefined as in related organizational configs, so that the right format is used but the placeholders are whatever they need them to be.

For extra-special configuration, we could defined INSTANCENAME_func that defaults to _get_INSTANCENAME so that a custom function can be used to do something completely different.

I think it is important to stipulate that the INSTANCENAME, however it is derived, should not be so dynamic that changes from run to run on the same platform/config and confuses the places we used it, like in naming the STATEFILE.

[jasonfleming]

Yes I totally agree that the INSTANCENAME must remain constant once it has been set and will therefore stay the same throughout the life of the instance. I can think of only one case where this might create unexpected side effects: when the Operator dynamically changes the meteorological forcing (e.g., from GFS to RRFS or from straight GAHM vortex to gfsBlend plus GAHM vortex, etc). This is something that asgsh would need to detect automatically to prevent it from happening. I think this limitation may actually be good because it distinguishes between different types of results that we don't want to mix up.

[jasonfleming]

I was thinking this feature would be implemented by adding a config/org_defaults.sh with a big case statement (similar to the mesh_defaults.sh file) based on the organizationID (which would be specified in .asgsh_profile or in the config file itself, similar to parameterPackage). The instance naming convention could be defined under a particular organizationID, perhaps taking into account an optional team ID or project ID.

The netcdf_metadata hash currently defined in config/model_defaults.sh could also be moved into the the case statement in the org_defaults.sh file so that the netcdf metadata is set properly.

[wwlwpd]

Okay, I think there's some context I do not yet know.