Everest Storage: Persistent storage for re-runs

[yngve-sk]

Currently, we assume an Everest experiment runs once, and is then done. When we store Everest results in storage on the file system, it may make sense to be able to re-run the optimization part, but not re-evaluate the forward model in ERT. Everest is currently not set up for this, but it could be.

Rough idea of how to complete this:
When opening an Everest config, if the optimization output directory already contains an ERT storage with an experiment that pertains to that specific config (discriminated by name of the config file for example), set that experiment as the "active storage" of the run of the EverestRunModel. Ideally, running the run model should run as normal, but will get simulator cache hits on all controls to evaluate, so only the optimization part is re-run (disregarding some file reading from ert storage)