Chrislab

* Remove requirements: notebook, ipynbname, ipywidgets, jupyterlab, jupyter, tornado==6.1

* Add/update members of LearningOption
* accuracy(): decorate with torch.no_grad()
* ProgressChecker:
- Add tb_logger, world_size, node_rank, local_rank, global_rank
- Remove epoch_per_step
* LearningOption:
- training_printing, checking_printing, checking_epochs, training_format, checking_format, testing_format, num_epochs, rate
- learning_rate, saving_policy, num_saving, check_rate_on_training, print_rate_on_training, print_rate_on_validate, print_rate_on_evaluate, print_step_on_training, print_step_on_validate, print_step_on_evaluate, tag_format_on_training, tag_format_on_validate, tag_format_on_evaluate
- saving_policy -> saving_mode
- name_format_on_saving
- seed -> random_seed
* ModelOption:
- home -> finetuning
* NsmcCorpus:
- log only when self.args.prog.local_rank == 0
* Update ClassificationDataset: corpus -> data
* Update dataframe(): minor change
* Use default_factory instead of default!
* Remove epoch_per_step
* batch_size -> train_batch, infer_batch
* Update ProjectEnv:
- Add _setup_logger(), set_output_home(), set_logging_file(), set_argument_file()
* Remove configure_csv_logger()
* Remove from_args()
* Remove set_seed()
* Remove load_arguments()
* Remove save_checkpoint()

* Update nlpbook.dp.corpus.py: import
* Update convert(): implement seq2seq-v4 style
* Update console_scripts
* Add ConvertApp
* Add EvaluateApp
* Update ConvertOption: Add s2s_type
* EvaluateApp + ConvertApp = CLI
* Not use NERCorpus.get_labels_from_data(): for changed data path
* Make flask_ngrok package and move flask_ngrok.py

* Update _convert_examples_to_cls_features():
- Make batch_text_or_text_pairs and use it
- Use args.data.num_check
* Update NsmcCorpus:
- Accept args and use it
- get_examples() -> read_raw_examples()
- _convert_examples_to_cls_features() -> raw_examples_to_encoded_examples()
* Update ClassificationDataset:
- Reject args
- Do not use cache
* Update TesterArguments:
- Add from_args()
* Sync task.py in cls and ner
* Do NOT call configure_dual_logger()!
* Update ServerArguments:
- Update __post_init__(): load from a dir also!
- Add from_args()

* Add convert_to_seq2seq_format() to nlpbook.ner.corpus
* Split klue-ner-v1.1_dev.seq2seq.tsv
* Add convert_to_seq2seq_format_v2()
* Add convert_to_seq2seq_format_v3()
* Update convert_to_seq2seq_format_v3()
* Add convert_to_seq2seq_format_v0()
* Add convert_to_seq2seq_format_v1()
* Add DPParsedExample, convert_to_seq2seq_format_v2()
* NER evaluation debugging mode!
* DP evaluation debugging mode!
* Remove chrislab.common
* Remove chrislab.language

* Move OptionData, ResultData, ArgumentGroupData, TimeChecker, CommonArguments to chrisbase.data
* Add MLArguments to nlpbook.arguments
* Move csv_logger from ProjectEnv to ProgressChecker
* Rename reconfigure_output() to configure_csv_logger()
* Move RuntimeChecking from nlpbook.arugment to chrisbase.data
* Move ArgumentsUsing from nlpbook.arguments to chrisbase.data
* Remove show() from MLArguments
* Update CommonArguments:
- Rename save_arguments() -> save_args()
- Rename info_arguments() -> info_args()
* Update JobTimer:
- Default verbose=True
- Default flush_sec=0.3
- self.args.info_args() called automatically
* Add MongoDB to chrisbase.util
* Update check_ip_addrs():
- Add param: mongo
* Update ProjectEnv:
- Add member: max_workers
* Update all_future_results: use tqdm
* Move EmptyTqdm, mute_tqdm_cls, time_tqdm_cls from chrislab.util to chrisbase.util
* Fix dataframe(): when member is None
* Update output_table(): use tqdm
* Adding find_fields(): for gathering