Implement HTTP request resilience

[paulolaup]

IS:

ATM, the is now uniform, project-wide handling of failing HTTP requests leading to unexpected workflow failures or continuations even though crucial parts of processes failed. There should be a configurable approach to handling such failures, including Retry and Repeat policies.

A fairly significant problem is posed by the limited processing queue of the terminology server we use. If multiple workflows are running in parallel they can quickly fill it leading to additional requests failing and ultimately the workflow run as well. Such occurrences should be handled by sensible Retry policies.

TODO:

1. Add retry logic to the HTTP clients we use, ideally with fine-grained error-handling (e.g. matching by status code, error message, etc.)
2. Externalize retry settings into project-specific config file