[9.0] Job parameters: use only OpenSearch

[fstagni]

BEGINRELEASENOTES

*WMS
CHANGE: only use Elastic/OpenSearch for storing JobParameters
CHANGE: ElasticJobParameters index new names: JobParameters_indexSplt

ENDRELEASENOTES

[arrabito]

I would have also a related but more general question about index prefix naming.
How do you manage the index prefix naming when you have 2 DIRAC instances (example a production one and a certification one) using the same OpenSearch service? I mean, do you give different index prefix names to distinguish between the 2 instances or do you use the same naming and 2 different tenants (I don't know if it's possible though).
I ask this here, because it can have an impact on this PR and it goes in the sense of make the index prefix name configurable also for job parameters.

[fstagni]

We normally use different OpenSearch instances for different DIRAC setups (same as we do not mix MySQL databases).

[andresailer]

Is that try/except for RuntimeError that was there before, really no longer needed?

[fstagni]

I think so.

[andresailer]

ElasticJobParametersDB still throws RuntimeError
e.g.:
https://github.com/DIRACGrid/DIRAC/pull/7292/files#diff-343b190022086de54ccf30f9a1ae0e7b1c5f5b312f95f48e9c91abf17f210dbaR43

[fstagni]

Then it's ObjectLoader().loadObject() that should catch that. Do you agree?

[andresailer]

No, the throwing comes from calling the c'tor of ElasticJobParametersDB in this line here, doesn't it?

        self.elasticJobParametersDB = result["Value"]()
                                                     ^^

[fstagni]

Correct. But, anyway, this is initialize() and it's OK to see the exception, if it happens.