Introduce proper framework metrics

[bitzl]

So far, Flusswerk defines two metrics out-of-the-box: total amount processed messages, and total time spent processing the message.

There are two issues with the current implementation:

1. the names are not following Prometheus best practices
2. framework metrics and user defined metrics share the same prefix as configured with flusswerk.monitoring.basename (default: "flusswerk")

This PR introduces two new metrics that are named independently from the value of basename and that follow best practices:

flusswerk_messages_total is the total number of messages processed since start of the application, flusswerk_processing_seconds the time spent on processing. Values from parallel threads are added up.