I found that a regular human gut metagenome sample needs around of 100G of space to store the intermediate results when running HUMAnN. In fact, it is completely unnecessary, because many files can be deleted in time. So, I create this PR to fix it. Hopely we can merge it. Thanks !