Automatically create data dir if missing before running harvester

This makes sense only for the harvester xtask as the others will not populate its contents, but doing it there will make it easier to start here, make this project more self-contained and simplifies our infrastructure automation.