Resume a partially done pipeline starts from the beginning.

[kemin711]

For this run, most jobs are done. Because of a typo in the time directive (used 'hr' instead of 'h' for Duration), my pipeline stopped. Then I fixed the typo. The resume the pipeline, instead of the run starting from where it is broken, it started from the very beginning. Even I used the -resume option. In this process, I used cache 'lenient', this should make the parser more flexible. Is there other settings or actions that I can take to reuse already computed results?