Single message processing optimizations

[cam-schultz]

Context and scope
Once #31 is implemented, individual Warp messages will be processed in their own goroutines, significantly increasing the overall application throughput when compared to the serial processing approach. However, there still remains opportunities to optimize the individual message processing jobs, targeting end-to-end single message latency. Some known examples include:

• Reducing the number of blocking RPC calls to API nodes
• Improving the locking logic on the inbound request handler singleton
• Consider processing incoming blocks async, to mitigate incoming block queue growth

#185 may provide insight into other bottlenecks that are good optimization targets. We should also explore tools such as flame graphs to identify hotspots.