This provides a model of how to do code sharing between the two
implementations we have, (with the expectation that the code in the
"turbot" package will be generic and that code in turbot_slack and
turbot_lambda will use that common code while doing whatever
additional marshalling is necessary for their particular application
environments).