Process cache has duplicate information

The index value is the same across system models and releases, but we treat each combination as independent, and therefore duplicate names, products, etc. leading to 40 MB of data when much less would be possible.

Start by tracing how the process cache represents combinations of system models and releases, then inspect where names and products are stored. Confirm how the shared index value is used and measure the cache size; done means removing duplicate information without changing the retrieved data.

Written by the indexing model from the issue text.