Fix ESM models export and add it to supported (#1328)

Author: eaidova

docs/source/openvino/models.mdx

@@ -49,6 +49,7 @@ Here is the list of the supported architectures :
 - DistilBert
 - Electra
 - Encoder Decoder
+- ESM
 - Exaone
 - Falcon
 - Flaubert

optimum/exporters/openvino/utils.py

@@ -369,6 +369,7 @@ def set_simplified_chat_template(ov_tokenizer_model, processor_chat_template=Non
     "deepseek",
     "deepseek-v2",
     "deepseek-v3",
+    "esm",
     "levit",
     "llama4",
 )

tests/openvino/test_modeling.py

@@ -1740,6 +1740,7 @@ class OVModelForMaskedLMIntegrationTest(unittest.TestCase):
         "deberta_v2",
         "distilbert",
         "electra",
+        "esm",
         "flaubert",
         "ibert",
         "mobilebert",

tests/openvino/utils_tests.py

@@ -63,6 +63,7 @@
     "donut-swin": "hf-internal-testing/tiny-random-DonutSwinModel",
     "detr": "hf-internal-testing/tiny-random-DetrModel",
     "electra": "hf-internal-testing/tiny-random-electra",
+    "esm": "hf-internal-testing/tiny-random-EsmModel",
     "exaone": "katuni4ka/tiny-random-exaone",
     "gemma": "fxmarty/tiny-random-GemmaForCausalLM",
     "gemma2": "katuni4ka/tiny-random-gemma2",