Allow skipping of doc loading

Sometimes we do repeated runs of the same workload against the same cluster, in such cases there is no need to load the documents so we shoul support skipping this step.

No files, tests, or entry points are named. Trace the document-loading step and the way repeated workloads are started, then identify how a skip option could be exposed; done means a repeated run can bypass document loading while normal runs still load documents.

Written by the indexing model from the issue text.