PostgreSQL Metrics Support

Nobody has claimed this yet.

### What problem are you trying to solve?

Deep inspection into PostgreSQL connections and the data being passed across them would provide more granularity in debugging performance issues for many database-driven applications.

### How should the problem be solved?

Implement a PostgreSQL codec that allows the Linkerd proxy to inspect the PostgreSQL wire protocol and provide metrics for the communications with the database server. This could be exposed as a CLI as well as providing Prometheus metrics.

Potentially useful metrics made available as a result of this:

- Open database connections
- Authentication failures/successes (more than a very few failures is indicative of either a very bad misconfiguration, or malice)
- Connection lifetime
- Query-response latency
- Response size

### Related Issues

As #1779, similar to #2214.

Start with related issues #1779 and #2214, then locate the Linkerd proxy's protocol-handling entry point and the CLI and Prometheus metrics entry points. Done means PostgreSQL wire-protocol inspection is implemented and the listed connection, authentication, lifetime, latency, and response-size metrics are exposed; the issue does not name specific files or tests.

Written by the indexing model from the issue text.