Fix cluster_id length check in startup.py

[MKoesters]

MAX_ID_LENGTH suggests a maximum length, not a static length. I adapted the condition accordingly

[XaverStiensmeier]

Good catch. At the beginning I planned to set it only as a maximum. Later I thought that it might be wise to not allow shorter cluster ids to make it less likely to have two clusters with the same id. However, this is not a fix. A fix is to perform a bibigrid.core.utility.id_generation.is_unique_cluster_id check on the given cluster id to guarantee that the id is not in use before creation, which is much safer.

changing != to > also has to be done for rest_startup.

[MKoesters]

Good catch. At the beginning I planned to set it only as a maximum. Later I thought that it might be wise to not allow shorter cluster ids to make it less likely to have two clusters with the same id. However, this is not a fix. A fix is to perform a bibigrid.core.utility.id_generation.is_unique_cluster_id check on the given cluster id to guarantee that the id is not in use before creation, which is much safer.

changing != to > also has to be done for rest_startup.

I'll have a look at this

[MKoesters]

@XaverStiensmeier I finally added the missing code
Does this look good to you?

[XaverStiensmeier]

I am on holidays starting tomorrow and will not have the time to thoroughly review it before.

Quick comment without actually testing it: Am I mistaken or is this preventing cluster termination, listing and update (so everything where cid is needed for a running cluster) as it now runs into

if not id_generation.is_unique_cluster_id(cluster_id, providers):
        msg = f"Cluster id ({cluster_id}) already exists"
        LOG.error(msg)
        raise RuntimeError(msg)

because the cluster to terminate of course already exists and the cid is checked independent of selected action. I might be mistaken as it is just my first thought without any testing.