Implement parametrisation for standardising model runs

[mo-nikosbaltas]

Background
Investigation ticket #315 identified limitations in the CMEW workflow where REF/EVAL handling was hard-coded, serialised, and not extensible beyond two runs. The investigation proposed a parameterised, run-agnostic workflow design enabling parallel execution and future extension to N runs.

Scope
Implement the investigated design in CMEW by:

• Introducing explicit task-level parameterisation over run
• Decoupling REF/EVAL logic from scripts and embedding it in the workflow graph
• Supporting per-run CDDS request generation and workflow execution
• Ensuring backward compatibility during transition where possible

Implementation details

• Add dataset as a Cylc task parameter (e.g. dataset=u-ab123, dataset=u-az513).
• Creation of the request and variables files should also be parallelised
• Parameterise:
• o configure_standardise<recipe,dataset>
• o standardise_model_data<dataset>
• Generate per-run CDDS request files (request_<dataset>.cfg) and variables files (variables_<dataset>.cfg)

Out of scope

• Changes to CDDS internals
• Anything involving model_runs.yml

Acceptance criteria

• Given SUITE_ID and REF_SUITE_ID template variables in rose-suite.conf
• When configure_standardise runs
• Then the request and variables files are made in parallel for each dataset
• And when standardise_mdoel_data runs
• Then two CDDS workflows run in parallel

Related tickets

• Investigation: #315

[mo-nikosbaltas]

Workflow for parallel ref/eval runs

[mo-nikosbaltas]

Added a third run:
"eval2": {
"model_id": "UKESM1-0-LL",
"suite_id": "u-az513",
"calendar": "360_day",
"variant_label": "r1i1p1f1"
}
The workflow (for parallel run) is shown below: