Consider compactifying example archives

[brookslogan]

Migrated from #64.

We expect compactification to be desired in most use cases, but currently have example archives that aren't compactified. Consider the following:

• Compactify example archive object archive_cases_dv_subset.
• Fix any tests that break due to referencing particular rows of archive_cases_dv_subset. (Maybe make them more unit-test-like with hand-crafted data, or maybe keep non-compactified data around in archive_cases_dv_subset_dt and construct a non-compactified archive directly from that DT.)
• Update vignettes that reference the old compactify=FALSE construction code for these example archives.
• Update benchmarking vignette if it was based on using the non-compactified data. (Maybe keep the non-compactified data around in archive_cases_dv_subset_dt, but compactify when forming archive_cases_dv_subset.)