The usage of the various caches in this package is mostly internal, and it is hard to figure out which ones are used, what they are used for, and how they can be configured.
On top of that there is no way to monitor the usage of the caches during the runtime of higher-level algorithms that use TensorKit.