This PR fixes some of the weirdness in exception handling:
Assumes that the user always wants a traceback, and remove all the--traceback shenanigans.
Server commands that throw error now return the full traceback as a Stacktrace message, or at least it tries to.
When the server side throws, the client reconstructs an exception (DruncServerSideError) from the Stacktrace, and, importantly, the message that was originally returned. This means that if one of the child controller throws after a command, the CLI client will still be able to understand what was ran on the other children (albeit not doing anything very interesting right now).
This PR fixes some of the weirdness in exception handling:
--tracebackshenanigans.Stacktracemessage, or at least it tries to.DruncServerSideError) from theStacktrace, and, importantly, the message that was originally returned. This means that if one of the child controller throws after a command, the CLI client will still be able to understand what was ran on the other children (albeit not doing anything very interesting right now).