Upgrade Error Logs Too Big

[jonwinton]

This feature made logs for upgrade errors (potentially) very large: #529

If you have too large of a JSON object, the JSON gets split onto two lines, which makes it harder to parse for tools like the Elastic Stack or Splunk.

The error stack is of known length and provides the location of the error but no context.

The component data could be very large and provides context, but without the direct pointer to where there was failure it might be harder to debug.

Other side note

We have no way of getting component data for an erroring upgrade without connecting directly to the db because all methods of egress for data will trigger an upgrase