Remove the head masking block in some vision models

[ydshieh]

What does this PR do?

#36545 introduced a mistake regarding the headmask

before that PR, we had some

        # Mask heads if we want to
        if head_mask is not None:
            attention_probs = attention_probs * head_mask

but after the PR, it became

    # Mask heads if we want to
    if attention_mask is not None:
        attn_weights = attn_weights * attention_mask

which is apparently wrong.

Furthermore, we already remove the support of head masking.

This PRs simply remove the blocks.

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: audio_spectrogram_transformer, deit, dinov2, dinov2_with_registers, dinov3_vit, dpt, ijepa, videomae, vit, vit_mae, vit_msn, vitpose_backbone, vivit, vjepa2, yolos

[vasqu]

Thx, what nasty bug 😓

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.