Standardize http.Client collector configurations

[alloveras]

Intent

The current implementation will use two different http.Client configurations depending on whether the collector is targetting ALL queues or only particular ones.

Problem

When targetting particular queues, the code will use http.DefaultClient and, as a consequence of its default values, it will use an infinite client timeout. This is particularly problematic as upstream servers may not have timeouts set either or use a very long one. Hence, when used in the Lambda environment, a bad behaving request can force the function execution time to skyrocket and go past the Lambda execution timeout.

Solution

To ensure that both configurations (all queues vs particular queues) do use the same HTTP client configurations with a sensible client timeout value.