Rename Plugin to AttentionAndDistributedPacking

[achew010]

Description

This PR renames the InstructLab plugin to AttentionAndDistributedPacking. Functionally, it is the same behaviour as the current plugin but all references of instructlab will be replaced.

export DATASET=benchmark_outputs/data/cache_mistralai_Mistral_7B_v0.1.json
python -m tuning.sft_trainer --model_name_or_path mistralai/Mistral-7B-v0.1 --padding_free huggingface --packing False --max_seq_len 4096 --learning_rate 2e-5 --torch_dtype float16 --use_flash_attn True --include_tokens_per_second True --num_train_epochs 1 --gradient_accumulation_steps 1 --gradient_checkpointing True --evaluation_strategy no --save_strategy no --weight_decay 0.01 --warmup_steps 10 --adam_epsilon 1e-4 --lr_scheduler_type linear --logging_strategy steps --logging_steps 10 --max_steps 100 --training_data_path $DATASET --per_device_train_batch_size 4 --output_dir benchmark_outputs/aadp --skip_memory_metrics False

[fabianlim]

we have renamed the ilab to aadp, we will be deleting the old pypi project https://pypi.org/manage/project/fms-acceleration-ilab/settings/. We will not delete the PYPI project right away, but will be done soon.