Treat PAR_Y similarly to other large projects in HCA

[kbergin]

See conversation in slack here: https://humancellatlas.slack.com/archives/C6C5S6B5W/p1550586092130300

Needs refinement

AC: We mimic the decisions of other large projects in our reference annotation

Open questions

• Would this require a reanalysis of all old data for both SS2 and Optimus?
• Do we want to do this?

┆Issue is synchronized with this Jira Dev Task