ActionHandler: better "life cycle" handling for execution actions occurring sequentially in the same process

[rsyring]

B/c we catch all exceptions in on_action, we need to manually send the exception to Sentry before logging it and returning: https://docs.sentry.io/product/sentry-basics/integrate-backend/capturing-errors/

Probably doesn't apply since we are sending it manually, but should look closer at the code to see what it does: https://docs.sentry.io/platforms/python/integrations/aws-lambda/manual-instrumentation/

Also, flask has certain events that are issued during a web request and those are used to init & cleanup each request. We should do those same things (e.g. clear the db session) when calling any action on the handler.