Implement index tracking to optimize list operations

[UnderscoreTud]

Problem

When adding an element to a list, Skript scans through the entire list to find an open spot for the element. This is a pretty slow operation with a time complexity of O(n), and adds up extremely fast, to the point where adding 10k elements to a list takes around 11 seconds on my machine.

Deleting a list using delete {_list::*} loops through every index and removes it individually. This is done to ensure all sublists are also deleted, but this is also done even if the list doesn't have any sublists!.
Also for some reason the logic for deleting a list is effectively this:

1. loop every index and delete its sublists
2. set the list to null (which does the same thing as above)

So, deleting a list actually deletes it twice

Solution

Adding

Use a custom TreeMap implementation that tracks the list's numerical indices instead of scanning the whole list every time something gets added. The map keeps track of the next open numerical index as elements are inserted and removed, so adding to a list no longer has to walk through every existing entry looking for a gap.

In the common case where the list indices are consecutive, adding an element just appends it to the end immediately. If there are gaps from deleted entries, the map reuses the lowest available numerical index instead. This brings list addition down from O(n) to effectively O(1) in the best case, while still avoiding the full scan that made the old design so slow.

With this implementation, adding 1 million elements to a list takes ~1.5 seconds on my machine, that's over a 70,000% improvement over the old design!

Deleting

Deleting a list now sets it to null, and the TreeMap implementation now keeps track of the sublist indices, so it only loops over the necessary indices rather than walking the entire list each time

Testing Completed

IndexTrackingTreeMapTest.java

Supporting Information

Benchmarks:

Benchmark	Before	After	Improvement%
Adding	11.5 seconds (10 thousand elements)	1.59 seconds (1 million elements)	72,327%
Setting	2.27 seconds	2.11 seconds	107% (Roughly equal, within margin of error)
Deleting entire list	1.07 seconds	0.08 seconds	1,337%

---

Completes: #891
Related: #891
AI assistance: Junie was used to generate the majority of test cases

[Absolutionism]

Solid 💪

[sovdeeth]

looking great