identify bottlenecks in code execution to speed up generation of sample(s)

[br-pki]

Output from a run of cProfile indicates slow processing in standard pandas calls (not pandorable?). See attachment
cProfile_profiling_results.txt

[br-pki]

Actually, output from line_profiler gives a more insightful, line-by-line breakdown of time. Biggest offenders are

• sample = take_sample(sentences, args.sample_type)
• these_sentences["Sentence_word_char_len"] = these_sentences.apply(
  lambda row: get_sentence_word_char_len(row), axis=1)
• sentences = get_sentences(min_sentences=args.minimum_sentences, the
  se_languages=args.languages)
• this_language = this_row["Language"]