It is a good suggestion. We have considered building retry settings into the API Call config before and will keep your specific suggestions in mind.
It should be possible to construct a retry mechanism, specifying each of these settings, using rules. A subset could be created on the API Request object, defining what represents a “Failure” using, for example, the “Response Code” and “Date Created” attribute values. Once a failure is 5 minutes old, act upon the related record to trigger the same API call again. Managing a maximum number of retries would need careful thought but is possible.
Thanks again for your suggestion,