`sort_custom` is comparing (unrepeated) element with itself

[MarianoGnu]

Tested versions

• Reproduced in 4.4.1

System information

Windows 11 - Vulkan 1.3.271 - Forward+ - Using Device Nº2: NVIDIA - NVIDIA GeForce RTX 3060 Laptop GPU

Issue description

I have a method that sorts some handlers by priority of execution, but noticed it was printing the warning i added to make sure 2 handlers didn't pick the same priority

func _input_handlers_compare(a: InputHandlerBase, b: InputHandlerBase) -> bool:
	if a.get_priority() == b.get_priority():
		push_warning("input handlers '%s' and '%s' share the same priority %f, will be sort alphabetically by file path" %
		[a.get_script().resource_path, b.get_script().resource_path, a.get_priority()])
		return a.get_script().resource_path > b.get_script().resource_path
	return a.get_priority() > b.get_priority()

after checking, i confirmed the element was not repeated in the array, sort_custom seems to just be comparing the element with itself as part of the algorithm, so i had to add an extra check as a workaround

func _input_handlers_compare(a: InputHandlerBase, b: InputHandlerBase) -> bool:
	if a.get_priority() == b.get_priority():
+		if a == b:
+			# sort_custom is just being dumb, no need to give a warning
+			return false
		push_warning("input handlers '%s' and '%s' share the same priority %f, will be sort alphabetically by file path" %
		[a.get_script().resource_path, b.get_script().resource_path, a.get_priority()])
		return a.get_script().resource_path > b.get_script().resource_path
	return a.get_priority() > b.get_priority()

Steps to reproduce

it's kinda explained in issue description

Minimal reproduction project (MRP)

Not provided (at least until i have time to sit and create one)

[kleonc]

Can confirm:

@tool
extends EditorScript

var same_count := 0

func _run() -> void:
	var array: Array = range(100)
	array.shuffle()
	array.sort_custom(func(a, b):
		if a == b:
			same_count += 1
		return a > b
	)
	print(same_count)

Example outputs (v4.4.1.stable.official [49a5bc7]):

14
12
9
10
15
8
10
10
11
8

I'm assuming that's a bug.