Refactor collector service for better error handling

The following discussions from !73557 (merged) should be addressed:

• @dzaporozhets started a discussion: (+5 comments)

  We rely on this attribute so we should require it

• @splattael started a discussion:

  Question (non-blocking) What are your thoughts on moving this rescue into CollectErrorService instead and let this service return a ServiceResponse (which can include a http_status)? This way we can simplify the API code and make sure that other callers of this service don't forget to catch this error.

  Edit: Again, non-blocking and we could easily follow-up if you agree 😅

• @splattael started a discussion:

  Question (non-blocking) Similar question to !73557 (diffs, comment 721781178). Should we move this check into the service? 🤔

• @splattael started a discussion:

  Question A record might be invalid due to several culprits:

  • Validation failed: Actor can't be blank
  • Validation failed: Description can't be blank

  Returning 400 Bad request is fine but I was wondering if we should keeping logging (in Sentry) those kind of errors and probably fix them later completely? 🤔