A trace chunk registry is needed by the consumer daemon since there is
no concrete representation of a tracing session object in those
processes.
The trace chunk registry allows the consumer daemon to match
newly-created streams to their current trace chunk. This integration
is performed in a follow-up commit.