Ability to load attributes on demand then to cache them instead of loading them on object instantiation #1509

What is your feature request?

After profiling my program that relies on python-plexapi, I found that _loadData was the function with the highest CPU time (~65% of the program's total CPU time).

This is largely because python-plexapi loads all available attributes on object instantiation. Most of which never get used. Instead, I suggest attributes get loaded on demand and then cached (with the cache key changing whenever the _data attribute changes).

For my use cases, where I only need access to the ratingKey, guid, viewCount, lastViewedAt, rating attributes. Using on demand access would cut down _loadData's runtime by over 80% in this scenario (according to my rough estimate from the flame graph below).

Pyspy profiler flame graph: