Fix expanding nodes with more than 32768 children

[sirgallifrey]

I created a test that reproduced the issue on #52 correctly expands node with 100.000 children
and the implementation fixed the issue.

Turns out that when chunking the nodes, the two arguments for splice (start and delete) were not taken into account and they add up from chunk to chunk making the subsequent start off by 2*orderPartsCursor(good thing I did the test with 100k nodes and not 40k)

One more thing, I ended up adding one extra test correctly collapses node with 100.000 children but since there was no issue with collapsing I didn't see this one fail. Let me know if you want me to remove it.

closes #52

[sirgallifrey]

I also tested on the storybook reproduction and it is working fine after this fix :)

[Lodin]

Hey @sirgallifrey, I finally found some time to get through that ticket again. I have a slightly smaller solution without orderPartsCursor variable in the fix/max-args-bug branch. Since it is very similar to your solution, and I also used your tests (with small updates), I would like to respect your contribution. So if you update that PR with the changes from the branch I mentioned above, I will merge the PR.

Sorry again for making you wait for so long.

[sirgallifrey]

@Lodin no need to be sorry, I understand you are donating you time for free to this project to the benefit of everyone. also the workaround I found allowed me to use it despite the problem :)

Thanks for allowing me to merge the PR, I appreciate it. Now I'm the one with little time, but probably I'll have some moments later this week to update it.

[sirgallifrey]

@Lodin can you have a look? I believe our branches are identical now

[Lodin]

LGTM