Numpy 2.0 compatibility

[jvansanten]

search_htm_ind caches the name of the index file and its contents in the same global dictionary, and checks whether the dictionary values contain a particular key. Since those values include the actual index contents, this compares the key to the index contents, which is a numpy array. On Numpy 1.2, this raised a warning and returned False:

/apps/catalogmatch/venv/lib/python3.9/site-packages/catsHTM/script.py:286: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison
  if VarName not in list(d.values()):

With Numpy 2.0, however, this is a hard error:

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

This PR replaces the global dictionary with a caching decorator, making the code simpler, faster, and compatible with Numpy 2.0.