10 more post-tuning OPs, regarding dialog data analysis from multiple aspects

[BeachWang]

Add 10 more post-tuning OPs, for fine-gained analysis of dialog data:

• 7 mapper ops:
  • LLM base:
    • dialog_intent_detection_mapper: Mapper to generate user's intent labels in feed back dialog data.
    • dialog_sentiment_detection_mapper: Mapper to generate user's sentiment labels in feed back dialog data.
    • dialog_sentiment_intensity_mapper: Mapper to predict user's sentiment intensity (from -5 to 5 in default
      prompt) in feed back dialog data.
    • dialog_topic_detection_mapper: Mapper to generate user's topic labels in feed back dialog data.
  • BERT base:
    • query_intent_detection_mapper: Mapper to predict user's Intent label in a query.
    • query_sentiment_detection_mapper: Mapper to predict user's sentiment label ('negative', 'neutral' and
      'positive') in a query.
    • query_topic_detection_mapper: Mapper to predict user's topic label in a query.
• 1 aggregator op:
  • Merge similar meta tags to one tag.
• 1 selector op:
  • tags_specified_field_selector: Select samples based on the tags of specified field.
• 1 grouper op:
  • naive_reverse_grouper: Split bathed sample to samples.

Others:

• Add MetaKeys for mapper op.
• Remove unneccesary lazy loader.
• Refine nested_set function.

[yxdyc]

LGTM