Reimplement objective scaling

[verveerpj]

Issue
This PR reimplements objective scaling using the new approach of ropt to transform variables and objectives between the model and optimization domains.

Approach
There are three commits:

1. A small commit that uses an improvement in ropt to simplify the storage code.
2. A refactoring of the forward model evaluator in the everest run model to introduce a clearer boundary between ropt and everest code.
3. The new implementation of objective scaling.

(Screenshot of new behavior in GUI if applicable)

• PR title captures the intent of the changes, and is fitting for release notes.
• Added appropriate release note label
• Commit history is consistent and clean, in line with the contribution guidelines.
• Make sure unit tests pass locally after every commit (git rebase -i main --exec 'just rapid-tests')

When applicable

• When there are user facing changes: Updated documentation
• New behavior or changes to existing untested code: Ensured that unit tests are added (See Ground Rules).
• Large PR: Prepare changes in small commits for more convenient review
• Bug fix: Add regression test for the bug
• Bug fix: Create Backport PR to latest release

[codspeed-hq]

CodSpeed Performance Report

Merging #9985 will not alter performance

_{Comparing reimplement-objective-scaling (7f42b79) with main (34f35d0)}

Summary

✅ 25 untouched benchmarks

[yngve-sk]

Could we pull out range(control_values.shape[0]) to a separate variable simulation_ids here for readability? Really it is a list of all ert realizations / sim_ids, I think pulling it out would clarify that too. Similarly (even though there is just one instance) within _run_forward_model

[verveerpj]

done.

[yngve-sk]

Hmm, it is more understandable I think, and could be revisited a few more times. In the long run we could maybe get rid of the -1 as evaluation id, maybe just use the ERT realization instead of it, since all evaluations are ultimately stored in ERT storage as a realization. This is future work either way

[verveerpj]

We are going to re-evaluate these ID's anyway. These ID's, and the batch-ID are passed, so you can map a result produced by ropt to a runpath on disk. So, if you have the batch and the model realizations, you look for the folder with the right simulation_# name, where # is the simulation ID. That is a bit primitive and cumbersome, we should think about improving this, and indeed ultimately there is some mapping to a location in ERT storage.

[yngve-sk]

LGTM