feat: nested selection conflict detection

[evpeople]

I implemented conflict path detection using HashMap. Using a prefix tree might also be a worthwhile option to consider.

Closes #688

[sxyazi]

Thank you for the PR!

Could you also add an insert_many(&mut self, urls: &[&Url], parent: Url) method? This is very useful for bulk selecting (selecting all) files - all of which have the same parent path, so it only needs to be detected once to improve performance.

Then insert() can simply be a wrapper for insert_many() like this:

	pub fn insert(&mut self, url: &Url) -> bool {
            self.insert_many(&[url], url.parent_url())
        }

[evpeople]

Ok, I will try it

[evpeople]

Considering that the paths corresponding to insert_many all have the same parent directory,
I implemented insert_many by modifying the logic in the original insert method that checks the object being inserted to instead check the first object in insert_many.

I didn't use the parent variable. I tried some methods that involved the parent variable, but they were not successful in passing the tests.

[evpeople]

To make the compilation pass, I added contains and get_inner methods. I am unsure if using the BTree inside inner within yazi-fm might lead to inconsistencies between HashMap and BTree.

[evpeople]

The usage of Deref here has taught me a lot.

[evpeople]

I have considered the issue of data overflow. Later, I realized that if the add operation is correct, then the count in the parent can safely be decremented to 0. If overflow occurs, it indicates the possibility of some concurrency-related modifications. At this point, the program should panic.

[sxyazi]

Hi, I made some little changes, and I believe it's all good now.

Thank you very much for contributing to this feature, let me merge it!