speed up rule policy contraction check

[wochinge]

Proposed changes:

• fix Including RulePolicy leads to infinite max_history being used to deduplicate training trackers #8295
• cache _rule_key_to_state as this a hotspot of the contradiction check according to the profiling results (viewable with snakeviz)
• this speeds up the RulePolicy training with Sara from 14.4 seconds to 5.26 seconds
• see memory considerations of the cache here
• follow up issue for bigger changes to the RulePolicy Improve RulePolicy prediction timings #8850

Status (please check what you already did):

• added some tests for the functionality
• updated the documentation
• updated the changelog (please check changelog for instructions)
• reformat files using black (please check Readme for instructions)

[samsucik]

Thanks @wochinge 😍