Improve metrics output for Auto Monitoring

With the current I2P demo app, it returns with 0ms latency and no errors. While this is great, it means that the monitoring charts essentially show zero with the exception of throughput and it's not very realistic for a normal app.

This change includes two new features:

1. It randomly sleeps for a period of time for each request. This introduces latency which can then be graphed.
2. It adds a new page /error which returns an HTTP 500 error. We cannot randomly generate errors on / as that is used as the health check and will circuit break the node. These errors can be manually generated as needed to show some of the monitoring capabilities off.

These changes strive to be as lightweight as possible to address these two needs.