* + add auto mode for analyzer: load all filters that produce stats to analyze the target dataset

* + add default mem_required for those model-based OPs

* - support wordcloud drawing for str or str list fields in stats
- support set the number of samples to be analyzed in auto mode. It's 1k in default.

* - take the minimum one of dataset length and auto num

* * update default export path

* * set version limit for wandb to avoid exception

* + add docs for auto mode

* + support t-test for Measure

* * fix some bugs

* - support analyze a dataset object
- optimize the logics of loading filters that produce stats and updating attributes of OPs

* - support analysis on tags in meta

* - support analysis with tagging OPs

* - move tags into the meta field

* - do not tell tags using their suffix
- suppress the error/exceptions in Monitor due to the termination of the main process
- exported stats file includes meta field in exporter

* - add insight mining

* * resolve the bugs when running insight mining in multiprocessing mode

* * update unittests

* * update unittests

* * update unittests

* * update readme for analyzer

* * use more detailed key

* + add reference