… for better reuse.

… order - the user of `acceptance_date` assumed the newest date would always be best - that's false logic.

… order - the user of `acceptance_date` assumed the newest date would always be best - that's false logic.

…ents, when looking for consent date after user's consent has been suspended (withdrawn)

…imelines

…imelines

…r consent deletion resulted in change

…s completed prior to withdrawal - was reporting status "withdrawn"

… `any_change_noted` check

…s "completion-date"

…e rebuilds for affected patients

Adherence report changes:
- always include a row for withdrawal, and use 'completed date' column
for date of withdrawal
- include completed visits even if they are completed after date of
withdrawal

qb_timeline changes:
- include rows after withdrawal for implementation of the above

migration changes:
- now that we must rebuild the timeline and adherence data for every
withdrawn user, it takes far too long to run and would render the
service unavailable until the migration completes. therefore, it only
removes the timelines and adherence data for that set of users, and will
rebuild when first requested.

NB: this PR merges into another branch, such that all may be a single
update on the develop branch.

---------

Co-authored-by: Justin McReynolds <mcjustin@uw.edu>
Co-authored-by: Ivan Cvitkovic <ivanc@uw.edu>