server: ReportTaskRun serialization failure perceived as TaskTimeout

[Snarr]

Bug

When the LittleHorse Server fails to serialize a ReportTaskRun request, it returns an LHApiException early, which means we don't forward a record through our Kafka Streams Topology reporting the state of the TaskRun. This means that the TaskRun state is usually perceived as a Timeout by the server, since no state is reported. We should forward a ReportTaskRun Deserialization Error or something equivalent as the state of the TaskRun to our topology.

Steps to reproduce

1. Register any TaskDef with the LittleHorse Server
2. Register any WfSpec that uses that TaskDef with the LittleHorse Server
3. Run a Task Worker for your TaskDef
4. Run the workflow and execute that Task
5. Your Task Worker should pick up the TaskRun via a PollTask request
6. Intercept the ReportTask RPC call before it is sent to the server by your Task Worker
7. Use an invalid json_obj value for the ReportTask.Output, such as a non-JSON formatted string "test"
8. The server will fail to serialize the ReportTask request and interpret the Task as timed out after the timeout period expires