Racecondition in jobgeneration may leave inconsistent job creation status

We have seen an example of lingering

Netarkivet error: Not creating jobs for harvestdefinition #69 (Kat1_dagligt_path_1level) as the previous scheduling is still running

it seems this is cause by the generation finishing around the time a new job generation has started. The adding/putting and removes to the maps in the Jobgenerator proberly needs to be protected better from race conditions.