[Feature request] Split setting 'Use existing Correspondents and Tags' into 2 selectable options

[stratiegery01]

In my prompt iteration to set up Paperless-NGX/AI before doing bulk uploads, I've noticed from my chatGPT logs that the token size for processing each document has been consistently increasing. I allow Paperless-AI to use preexisting values for consistency sake, but I also try to have a set number of document types and tags, whereas I allow any amount of correspondents to be used or generated (think of any library with millions of different authors, but only a set number of aisles/categories of which to search).

It would be helpful to reduce the system prompt during each document processing request, lowering the number of tokens and thus speeding up the process. Therefore I request that this 'advanced setting' option for "Use existing Correspondents and Tags?" be split into 2 different settings options, allowing more flexibility in prompt refinement.