- GLiNER evaluation for all datasets except pile, with results (which are invalid for figer because it currently only returns one annotated entity for each sentence)
- fixed test_datasets inaccurate number of sentences error due to recent changes regarding limiting numbers of outputted instances

Figer: Save preprocessed data into files, download&extract dataset automatically, make fine/coarse have same sentences

- Dataset instances for Pile-NER-type and CoNLL now reproducibly random in database manager
- added GLiNER to common interface
- tested GLiNER on 100 instances
- fixed division by zero error

Added example output csv and txt logs;
The LLM experiment now has its own folder in experiments

Working pile-NER-type dataset with common dataset interface in data_manager and testcase in test_Datasets