Client V2 must accept config file (at lest in development mode)

The best way to get clients running from a standalone config is via changes to the v2-client, since otherwise we would need to bypass large chunks of control code to get a "standalone mode"

Two non-exclusive ways to do this:
- Have the connection info for the agent be read over a config file and not stdin
- Have the client accept a config file with a hard-coded config, and treat that config like it's coming from an elastic-agent server