Add HTTP proxy support to Agent Cluster

Current Situation

• If a Proxy Server is in place for access from a Master to an Agent, for example using the NginX or Apache (httpd) web servers, then this connection works fine as long as the Agent is available. 
• If the Agent becomes unavailable then the Proxy Server reports HTTP status code 502 "bad gateway" which is considered by the Master as a job error not as an indication for unavailability of the Agent. As a consequence the Master will not fail-over execution of jobs to the next Agent in the Agent Cluster.

Desired Behavior

• The Master should check for specific HTTP status codes 500 and higher to determine that the Agent is unavailable and to fail-over execution to the next Agent.