Move failed account takeover logging to redshift, include error

[maddiedierker]

Builds off of Redshift logging in #24720.

Currently, when an account takeover fails due to an attempt to takeover an account with progress, we log the failure to Honeybadger (example). We originally thought that this wasn't ever really happening, but realized we were wrong. This meant lots of Honeybadger notifications for this use case.

Instead, we should be logging this to Redshift, leveraging the current logging in place. This adds a new event to Redshift:

Study: user-soft-delete-audit
Events: cancelled-silent-account-takeover / cancelled-oauth-account-takeover

• user_id: user ID of account being taken over
• data_int: user ID of the account taking over user_id's account
• data_string: OAuth provider involved in the account takeover
• data_json:
  • user_type
  • error (currently always "Attempted takeover for account with progress.")

[maddiedierker]

move HB takeover logging to Redshift

[islemaster]

👍 Looks good. It'll be interesting to see if we can eliminate this case with later refactoring!

[ewjordan]

LGTM, once the CircleCI failures are resolved - looks like there are some related ones, where we used to test for events going to Honeybadger. Also some PD-related failures, which I doubt are related.