One of the aim of those Splunk HF is to offload the Splunk Indexer on parsing Pipeline, Merging Pipeline and Typing Pipeline. Due to that the data coming from Splunk HF are already "processed" and our Indexer are mostly processing them only in the Index Pipeline.
How many source systems, HFs and indexers you have? Probably more interesting is how well your events are distributed over indexers than how well those cores/pipelines are used in any particular moment. Here is excellent tools to check this https://github.com/silkyrich/cluster_health_tools.