This PR introduces a cache that allows the scanner to avoid emitting multiple requests to verify the same credential. In practice, it doesn't seem to reduce scan time at all, but it does seem to reduce the number of calls to FromData rather drastically.
The cache is implemented as an opt-out feature that can be disabled with a new CLI flag. If we don't like this, we can change it.
The metrics collection hopefully isn't too architecture-astronauty; I wanted to create something useful here that could also accommodate future Prometheus configuration without making the implementation all stupid.
