Improve quality of --small test run

[gailin-p]

Improvements to make --small test run option more useful

--small is meant to be a quick way to exercise the entire data generation pipeline and validation checks, so during development we can be confident that we're not breaking anything. Some improvements would make it easier to run:

• [ ] Small should be run as a pre-commit hook so that the pipeline always works
• [ ] With the --small option, the partial_cems dataframe has no entries. We want to make sure the --small run has some partial cems data, so that parts of the code dealing with scaling and handling partial cems data are exercised by --small. (issue identified in #162 )