Non-fault policies/strategies (e.g. polling)

[kmcclellan]

Any thought given to expanding this library to express execution policies/strategies that don't involve exceptions? A simple example is a poller that runs periodically using a random decorrelation coefficient. I've found myself wanting to leverage the configurable "backoff" delay + random jitter implemented in this library in scenarios outside the realm of fault handling.

[martincostello]

This is supported via methods such as HandleResult<T> as documented in the README.

For example, you could retry until the result of a method was true.