support for arbitrary health checks

[sawall]

Operators of the service may wish to add additional endpoints to health check to roll up into the picture of signal node health. This can be expressed as a JSON string in environment variables, something like:

EXTRA_SERVICE_CHECKS={
    [
        {
            "name": "[NAME]",
            "url": "[ENDPOINT]",
            "include_contents": [boolean],
            "affect_global": [boolean]
        }, [...]
    ]
}

The URL is expected to return 200 if healthy. Any other response or failure to reach the endpoint will result in it being considered unhealthy.
The outcome of this check will be added to the "services" section of the signal report. If "name" is "prosody_egress", then:

    "services": {
        "prosody_egressHealthy": [boolean],
        "prosody_egressReachable": [boolean],
        "prosody_egressStatusCode": [int], ...
        (OPTIONAL): "prosody_egressStatusContents": [stuff]
    }

If affect_global is true, then the result of health checking this service will affect the "healthy" status in /signal/report and whether /signal/health returns a 200.

[sawall]

This should be a refactor of all of health_collector.ts to be more generic.