LP#1729620: (follow-up) do not suggest timestamp manipulation

LP#1729620: (follow-up) do not suggest timestamp manipulation

The suggestion in the upgrade script and the release notes
to consider adding triggers to update biblio.record_entry.timestamp
upon updates of linked call numbers or items is not a good one:

[1] implementing it means that any workflows that require knowing
    when a bib itself has changed become untrustworthy
[2] it would cause tremendous bloating of auditor.biblio_record_entry_history,
    as every single change to an item -- including ones caused by
    circulation activity -- would result in a new entry.
[3] item changes due to circulation activity aren't necessarily
    going to want to trigger record updates by the harvester

Wanting the update timestamp for OAI harvest to reflect changes
to items is a perfectly valid use case, but this is not the way.

Signed-off-by: Galen Charlton <gmc@equinoxOLI.org>