Feature caching multiple runs

Feature Caching is a great addition to FME that gives the user a much better ability to inspect what happens inside of a workspace. One addition that I would like to see to feature caching is the ability to cache multiple runs/iterations from a transformer. This would be of a help for iterative workflows when creating a workspace. For instance, I'm working with OCR-reading and rasterization techniques at the moment trying to find the parameters that gives the best OCR-reading result from tesseract. A task that is far from clear cut and that relies heavily on trial and error. Having the ability to cache multiple runs would help this iterative workflow a lot in comparing which changes give the best result. Ideally multiple-run caching would be set on a transformer level, as to not exponentially increase the amount of memory that is used. With a parameter that specifies the amount of runs to cache.