exist time gap for computing for the same dataset

[odango16303]

Dear authors,

Thank you for your useful program.
I found something strange.
For certain datasets, the computation of PH take much shorter if the indexing of points is reordered.
Here are some examples with my comments.
https://github.com/odango16303/elapsed-time-lag-for-same-datasets/blob/master/elapsed%20time%20lag%20for%20the%20same%20dataset.ipynb

I would appreciate if you could find the reason.

Respectfully,
Katsuhiro Oda