feature: Make the algorythm work On Demand #4
Description
Current Behavior
The current stages of generating a vim-dan documentation , when dealing with online content (possibly .html) , has to follow all the steps, but the files have to be stored in the local system on the Indexing Stage , so the Parsing Stage and the Writting Stage works on those files indexed.
In order to achieve LINKS-TARGET-INDEXING , the Parsing Stage needs to be separated from the Writting Stage , in the Parsing Stage there is a first stage which is currently called LINKS-INDEXING which makes a .csv with all the html elements with ids which can become link targets. Then it iterates again on INUSE-LINKS-INDEXING to modify that .csv ./links-parsed , to populate the column of in_use , in order for then later on the Writting Stage write the minimum amount possible of inline link targets , cause otherwise every html element with an id could be a link target , but not all of them are been used , so placing <I=3f#8t>' etc... all over the .dan` file would be cluttering it.
Those stages are necessary to be separated with the current state of the algorythm (by using the pandoc workflow).
Intended behaviour
In order for vim-dan-generator to become as automated as possible, the best would be to delete Stages of the process , also simplifying the specification of the rule-set per documentation, to the ultimate point in which even no parameters had to be inputed , just a target website, and with the help of LLM's may identify the Pages of Interest, its selectors, Parsing on Demand , Writting on Demand.
On Demand Parsing and Writting
By On-demand parsing and Writting , would be to eliminate the Indexing Stage performing the Parsing Stage and the Writting Stage in a per file basis without storing.