Mailer jobs fail when we can't find a record
In !59907 (merged), I set the default Sidekiq retries back to 25 from 3 for all new workers. One side effect that I didn't anticipate - but should have - is that this also applied to mailers, as we don't define the worker class for those in our application code.
From https://log.gprd.gitlab.net/goto/dacf6ee33f1b715a4dae85eac2f55cd0 and https://log.gprd.gitlab.net/goto/746a6a6b29359da541b4e93175461e22 we can see that some mailers persistently fail due to being unable to find an associated database record. For instance, we schedule a mail about a note someone has added, but they delete the note before Sidekiq picks up the job.
For those cases, we can probably just ignore the failure and not retry the job.