[EXPERIMENT][WIP] Add SmolLM3 support for OpenVINO export

[mlukasze]

⚠️ AUTOMATICALLY GENERATED BY OMEGA AGENT — REQUIRES HUMAN REVIEW ⚠️
This PR was created by an AI agent as part of automated model enablement.
A human maintainer must review and approve it before it can be considered for merge.
Do NOT merge without human review and sign-off.

Summary

This PR adds OpenVINO export and inference support for SmolLM3 (HuggingFaceTB/SmolLM3-3B), and fixes a bug in OVModelForCausalLM that caused generation drift with recent transformers (>=5.x).

Changes

1. Add SmolLM3 config (optimum/exporters/openvino/model_configs.py)

Added SmolLM3OpenVINOConfig for OpenVINO export support:

• Inherits from TextDecoderWithPositionIdsOnnxConfig (position_ids support)
• Uses MistralDummyPastKeyValuesGenerator (GQA: 4 KV heads vs 16 attention heads)
• Uses OVDecoderModelPatcher for SDPA patching

Without this change, export fails with:

sdpa_mask_without_vmap() missing 1 required positional argument: 'cache_position'\n```\n\n### 2. Fix position_ids double-append (`optimum/intel/openvino/modeling_decoder.py`)\n`_update_model_kwargs_for_generation()` was double-appending `position_ids` with transformers 5.x, shifting token positions during autoregressive decoding and causing generation drift.\n\nFixed by removing the extra append and relying on transformers `GenerationMixin` update logic.\n\n### 3. Regression test (`tests/openvino/test_decoder.py`)\nAdded test to verify `OVModelForCausalLM` does not mutate position_ids beyond the superclass result.\n\n## WWB Accuracy Results\n\n| Format | Score | Deviation | Status |\n|--------|-------|-----------|--------|\n| FP32 CPU | 1.0000 | 0.00% | ✅ PASS |\n| INT8 CPU | 0.9651 | 3.49% | ✅ PASS |\n| INT4 CPU | 0.9398 | 6.02% | ❌ FAIL (above 5% threshold) |\n\n- FP32 achieves perfect score after fix (was 12.67% deviation before)\n- INT4 quantization loss is inherent; not related to the decoder bug\n\n## Related\n- GitHub issue: intel-sandbox/applications.ai.openvino.meat#912\n- Jira: CVS-183434

[mlukasze]

Superseded by upstream PR: huggingface#1761