Catch timeout errors from API

[ehenneken]

This part of the code catches exceptions that have not been caught

        try:
            return solve_for_fields(hypothesis)
        except Undecidable as ex:
            possible_solutions.extend(ex.considered_solutions)
            reason = ex.reason
        except (NoSolution, OverflowOrNone) as ex:
            current_app.logger.debug("(%s)"%ex.__class__.__name__)
        except (Solr, KeyboardInterrupt):
            raise
        except Exception as ex:
            current_app.logger.error("Unhandled exception of type {0} occurred with arguments:{1!r}, thus killing a single hypothesis.".format(type(ex).__name__, ex.args))
            current_app.logger.error(traceback.format_exc())

So, one exception it does not catch is a ReadTimeoutError (which happens when a request takes too long and the API closes the connection).

An addition except ReadTimeoutError rule should be added, so that only unexpected exceptions will get caught by the final step.