Investigate best way to handle transcoding errors

[danillouz]

Currently errors are captured and send to the serverless dashboard, but it would be good to also "flag" the recording item in the DB.

[danillouz]

We could update the "status" of the corresponding recording item, so the user knows processing an update failed.
A good approach is usually by using a dedicated SNS topic for errors.

Additionally, if we'd want to retry a transcoding job, we should us a Dead Letter Queue (DLQ), but if we want to explore adding this functionality, we should open a separate issue, because it comes with its own challenges.

[danillouz]

AWS recently announced Lambda Destinations which seems interesting and convenient to deal with errors.