Add possibility to skip special tokens during attribution

[gsarti]

Description

This PR addresses the possibility of skipping special tokens during attribution using the skip_special_tokens=True argument.

Example usage (regular attribution):

import inseq

model = inseq.load_model("mymusise/CPM-Generate-distill", "integrated_gradients")

out = model.attribute(
    input_texts=["她们是飞行员还是制片人", "她们是飞行员还是制片人"],
    generated_texts=["她们是飞行员还是制片人？", "她们是飞行员还是制片人吗？"],
    step_scores=["probability"],
    skip_special_tokens=True,
)

out.show()

Example usage (contrastive attribution):

import inseq

model = inseq.load_model("mymusise/CPM-Generate-distill", "saliency")

out = model.attribute(
    input_texts="她们是飞行员还是制片人",
    generated_texts="她们是飞行员还是制片人？",
    skip_special_tokens=True, # New argument to skip special tokens
    contrast_targets="她们是飞行员还是制片人吗",
    attributed_fn="contrast_prob_diff",
    step_scores=["probability", "contrast_prob_diff"],
)

out.show()