[Fix] label_list not being set for NLP token classification training if distillation teacher and student labels do not match

[markurtz]

Fix for: https://app.asana.com/0/1204070232568744/1204101088568500/f

Testing:
Rerunning the procedure described here and ensuring convergence: https://colab.research.google.com/drive/1WuWJMYY-_S-JP711bLYSRxBbcb66X1ft

Rerunning the following example and ensuring there isn't a crash:

sparseml.transformers.train.token_classification \
  --model_name_or_path zoo:nlp/masked_language_modeling/obert-base/pytorch/huggingface/wikipedia_bookcorpus/pruned90-none \
  --recipe zoo:nlp/token_classification/obert-base/pytorch/huggingface/conll2003/pruned90_quant-none \
  --distill_teacher zoo:nlp/token_classification/obert-base/pytorch/huggingface/conll2003/base-none \
  --dataset_name conll2003 \
  --output_dir sparse_bert-token_classification_conll2003 \
  --per_device_train_batch_size 32 --per_device_eval_batch_size 32 --preprocessing_num_workers 6 \
  --do_train --do_eval --evaluation_strategy epoch --fp16 --seed 29204  \
  --save_strategy epoch --save_total_limit 1

[dbogunowicz]

LGTM

[bfineran]

LGTM pending summary of further testing

[bfineran]

blocking land due to some issues raised during testing

[bfineran]

fixes added to ensure plaintext string labels are preserved on fix - the only time pre 1.4.1 behavior is overwritten is when the original bug condition is met

[bfineran]

verified convergence and string label names are preserved for sparsification training runs with:

• conll from HF datasets (w/ distillation)
• conll from local (w/ distillation)
• wnut from HF datasets (no distillation)
• wnut from local (no distillation)