Connect timeout to the collector should be higher

[mattbaker]

Other agents have an HTTP connect timeout as high as 2 minutes when they connect to the new relic collector. We should increase the timeout the agent uses for contacting the collector specifically.

Users can get around this by setting custom httpc options but that value is global and there could be cases where it causes a GenServer.call to timeout if another process is waiting a long time on an http connection and doesn't respond quickly enough.

Additional thoughts

• The higher timeout should be specific to the collector code (for now)
• We'll need to analyze how a higher timeout impacts other parts of the application. For example, if it causes a given genserver process to not response to messages while it waits then other parts of the agent may experience timeouts in their own calls to the process. This could lead to some weird bugs. The solution will need to include analyzing what other parts of the system may need to change to account for a higher connect timeout.