merge dict_word_tokenize() and word_tokenize()

[bact]

• Reduce duplicated code.
• dict_word_tokenize() is a word_tokenize() with a custom_dict argument.
• dict_word_tokenize() is currently kept (inside is just a call to word_tokenize()) but should be deprecated
• word_tokenize() argument whitespaces is now keep_whitespace to make is more explicit, default behavior is to keep whitespaces
• fix word_tokenize() call in FrequencySummarize to properly pass tokenizer engine value

[coveralls]

Coverage increased (+0.2%) to 81.382% when pulling 7671f0e on bact:dev into 94dd259 on PyThaiNLP:dev.