[EXPERIMENT][WIP][OpenVINO] Add support for Nomic Bert

[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.

---

Known blocker requiring human action before merge: the Hub-hosted tiny test fixture optimum-intel-internal-testing/tiny-random-nomic-bert referenced in the new utils_tests.py entry does not exist yet. A creation script (create_tiny_nomic_bert.py) has been generated and needs a maintainer with write access to the optimum-intel-internal-testing org to run it and upload the result before the new CI test can pass end-to-end on the Hub-hosted path. The new tests were independently verified passing (2 passed) against an equivalent local tiny native-transformers nomic_bert model in this session (substitution reverted before commit; committed diff still points to the pending Hub path).

What does this PR do?

Adds OpenVINO export and inference support for nomic_bert (NomicBertOpenVINOConfig), enabling nomic-ai/nomic-embed-text-v1.5 and other native-transformers nomic_bert architecture models.

• Added NomicBertOpenVINOConfig as a one-line subclass of BertOpenVINOConfig in optimum/exporters/openvino/model_configs.py, following the same minimal pattern used by sibling BERT-family models (Albert, ConvBert, Electra, RoFormer, SqueezeBert, MobileBert, RemBert). No custom ModelPatcher is required; real-model tracing succeeds cleanly.
• Added "nomic_bert" to OVModelForFeatureExtractionIntegrationTest.SUPPORTED_ARCHITECTURES in tests/openvino/test_modeling.py.
• Added the tiny-model registry entry to tests/openvino/utils_tests.py.
• Added a Nomic Bert row to docs/source/openvino/models.mdx.

Verified with a real end-to-end export of nomic-ai/nomic-embed-text-v1.5 via optimum-cli (single-file openvino_model.xml/.bin layout, no fatal warnings) and OVModelForFeatureExtraction inference matching expected shape/dtype for both PyTorch and NumPy inputs. WWB accuracy validation passed on both CPU (0.9999992) and GPU (0.9999988).

Installation instructions

pip install git+https://github.com/mlukasze/optimum-intel.git@enable/nomic-ai-nomic-embed-text-v1.5
pip install --pre -U openvino openvino-tokenizers nncf --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly

Exporting cmd-line

optimum-cli export openvino --model nomic-ai/nomic-embed-text-v1.5 --task feature-extraction ov_model

Inference script

from optimum.intel import OVModelForFeatureExtraction
from transformers import AutoTokenizer

model_id = "nomic-ai/nomic-embed-text-v1.5"
model = OVModelForFeatureExtraction.from_pretrained("ov_model")
tokenizer = AutoTokenizer.from_pretrained(model_id)

inputs = tokenizer("search_query: What is TSP?", return_tensors="pt")
outputs = model(**inputs)
print(outputs.last_hidden_state.shape)

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[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.