allow backfilling of counts and rpkm without triggering preprocessing

[ppavlidis]

To address curation issue 362.

By using the the noLog2cpm option for RNASeqDataAddCli should allow us to backfill counts and rpkm for old data sets in a minimally-invasive way.

Needs testing.

TODO

• [x] unit test for the CLI tool and argument parsing
• [x] unit test for skipping log2cpm recomputation and post-processing
• [ ] test case when library size differs

[arteymix]

Ideally we would just compare the log2cpm with the new one and trigger post-processing if they differ.

I'll write some unit tests for the CLI.

[arteymix]

I'll write unit tests instead of adding an integration test because it's impractical to check if the preprocessing service has been interacted with.