Interrupting and re-running recurring reports can cause duplicated schedule entries

Here's a branch that addresses this by adding a unique index on the report, folder, runner, run_time, and email fields of reporter.schedule: https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/jboyer/lp1893463_uniq_recurrence / working/user/jboyer/lp1893463_uniq_recurrence

Clark is also taught that the specific error caused by this constraint is not a problem.

Dupe? #1892535

Works well for us! Pushed to master, rel_3_5 and rel_3_6. Could easily be backported to rel_3_4.

Follow-up commit, addressing the need for de-duplicating the reporter.schedule table before adding the index, also adding release notes, available for review at the tip of https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/csharp/lp1893463_upgrade_script_improvements

I have added a signoff to Chris Sharp's improved upgrade script in user/dyrcona/lp1893462_upgrade_script_improvements_signoff on the working repo:

https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/dyrcona/lp1893462_upgrade_script_improvements_signoff

I went ahead and pushed the signed-off follow-up to rel_3_5, rel_3_6, and master. Thanks, Jason!

I've pushed a branch to address an issue with the index and upgrade script. From the commit message:

Unique indexes on nullable columns will allow multiple conceptually unique rows if the nullable columns are, in fact, NULL because NULL does not equal itself. This commit uses COALESCE to make sure that the nullable email column in the reporter.schedule table gets a value of the empty string for the purposes of the unique index. The upgrade script now also takes this into account and ignores the email column.

Branch at:

https://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/miker/lp1893462_upgrade_script_improvements_signoff

Given the recent update to the update script and the fact that the point of this is to delete data, I've elected to bump this to 3.6.1, with this excluded from 3.6.0.

Thanks, Mike - I pushed your commit to master.

Mike's follow-up needs to be committed to 3.5.

OK, follow-up is in rel_3_5 now.