auto_compute_allocation enhancements - Adds accelerator hours, slurm attrs, slurm account and fairshare naming...

[ds-04]

This PR enhances the auto_compute_allocation plugin with the following changes:

• Ability to add accelerator hours (gauge) for both training and non-training projects
• Ability to add slurm attributes via ENV tuple for both training and non-training projects
• Control slurm account naming via a string
• Some limited control of institution fairshare naming (similar to slurm account naming) via a string
• README.md updates and improvements
• Consistent use of the word institution rather than institute

---

N.B. there is overlap between the variables that can be used to define the custom strings for:

AUTO_COMPUTE_ALLOCATION_FAIRSHARE_INSTITUTION_NAME_FORMAT
AUTO_COMPUTE_ALLOCATION_SLURM_ACCOUNT_NAME_FORMAT

... this is intentional and provides ease of maintenance - by using two separate files/functions to perform these.

---

tagging @cecilialau6776

[cecilialau6776]

Also, I don't see a particular reason to put fairshare_institution_name and slurm_account_name in their own files instead of in utils.py. Looks good otherwise

[cecilialau6776]

missing _TRAINING?

[ds-04]

thanks will amend README to add _TRAINING

the separate files were used for ease of maintenance and to avoid a bloat of the utils.

[ds-04]

Should now be ready - minor changes were made it the 2x force pushes ...

1. _TRAINING in README
2. string not dict in comment

[ds-04]

there's possibility a problem with a config/env var ... I'm working on establishing if that definitely is the case

(an import error)

[ds-04]

@aebruno i think this is ready now, the last force commit fixed the error in the plugin's defaults ... a typo had crept in