phm version 2.1.0 (2025-09-30)

- Major performance improvements: many core functions rewritten in C++ for substantial speedups.
  - Obtaining data from PubMed has been rewritten in C++ and is now much faster.
  - Phrase-document parsing and phrase collection are now much faster and memory-efficient.
  - Text clustering functions now use optimized sparse matrix representations and C++ centroids.
  - Pairwise and matrix text distance computations have been re-implemented in C++ for faster execution.
- Minor bug fixes and code cleanup:
  - Punctuation inside words are no longer used to separate phrases.
  - Braces only separate phrases if their contents contain spaces; single words inside braces are considered part of a phrase.
