Update de semchunk for the chunking methods

### Requested feature

While trying to determine an appropriate chunk size, I noticed that the current solution does not include a basic token overlap feature. This feature allows for a more contextualized partitioning of data in cases where a whole section cannot be fully included.

I understand that some users may not be in favor of overlapping elements because they can negatively impact data retrieval from a vector database. Depending on the overlap size and the retrieval technique used, this could lead to multiple instances of the same content being fetched.

### Alternatives

I have observed that the latest version of **semchunk** already includes overlap support and other useful features, in addition to improving the chunking mechanism.