feat: Add suffix to openai_completions (#2449)

For code completion apps need "fill in the middle" capabilities. 
Added option of `suffix` to `openai_completion` to enable this. 
Updated ollama provider to showcase the same. 

### Test Plan 
```
pytest -sv --stack-config="inference=ollama"  tests/integration/inference/test_openai_completion.py --text-model qwen2.5-coder:1.5b -k test_openai_completion_non_streaming_suffix
```

### OpenAI Sample script
```
from openai import OpenAI

client = OpenAI(base_url="http://localhost:8321/v1/openai/v1")

response = client.completions.create(
    model="qwen2.5-coder:1.5b",
    prompt="The capital of ",
    suffix="is Paris.",
    max_tokens=10,
)

print(response.choices[0].text)
``` 
### Output
```
France is ____.

To answer this question, we 
```

tests/integration/inference/test_openai_completion.py

@@ -22,9 +22,6 @@ def provider_from_model(client_with_models, model_id):
 
 
 def skip_if_model_doesnt_support_openai_completion(client_with_models, model_id):
-    if isinstance(client_with_models, LlamaStackAsLibraryClient):
-        pytest.skip("OpenAI completions are not supported when testing with library client yet.")
-
     provider = provider_from_model(client_with_models, model_id)
     if provider.provider_type in (
         "inline::meta-reference",
@@ -44,6 +41,23 @@ def skip_if_model_doesnt_support_openai_completion(client_with_models, model_id)
         pytest.skip(f"Model {model_id} hosted by {provider.provider_type} doesn't support OpenAI completions.")
 
 
+def skip_if_model_doesnt_support_suffix(client_with_models, model_id):
+    # To test `fim` ( fill in the middle ) completion, we need to use a model that supports suffix.
+    # Use this to specifically test this API functionality.
+
+    # pytest -sv --stack-config="inference=ollama" \
+    # tests/integration/inference/test_openai_completion.py \
+    # --text-model qwen2.5-coder:1.5b \
+    # -k test_openai_completion_non_streaming_suffix
+
+    if model_id != "qwen2.5-coder:1.5b":
+        pytest.skip(f"Suffix is not supported for the model: {model_id}.")
+
+    provider = provider_from_model(client_with_models, model_id)
+    if provider.provider_type != "remote::ollama":
+        pytest.skip(f"Provider {provider.provider_type} doesn't support suffix.")
+
+
 def skip_if_model_doesnt_support_openai_chat_completion(client_with_models, model_id):
     if isinstance(client_with_models, LlamaStackAsLibraryClient):
         pytest.skip("OpenAI chat completions are not supported when testing with library client yet.")
@@ -102,6 +116,32 @@ def test_openai_completion_non_streaming(llama_stack_client, client_with_models,
     assert len(choice.text) > 10
 
 
+@pytest.mark.parametrize(
+    "test_case",
+    [
+        "inference:completion:suffix",
+    ],
+)
+def test_openai_completion_non_streaming_suffix(llama_stack_client, client_with_models, text_model_id, test_case):
+    skip_if_model_doesnt_support_openai_completion(client_with_models, text_model_id)
+    skip_if_model_doesnt_support_suffix(client_with_models, text_model_id)
+    tc = TestCase(test_case)
+
+    # ollama needs more verbose prompting for some reason here...
+    response = llama_stack_client.completions.create(
+        model=text_model_id,
+        prompt=tc["content"],
+        stream=False,
+        suffix=tc["suffix"],
+        max_tokens=10,
+    )
+
+    assert len(response.choices) > 0
+    choice = response.choices[0]
+    assert len(choice.text) > 5
+    assert "france" in choice.text.lower()
+
+
 @pytest.mark.parametrize(
     "test_case",
     [

tests/integration/test_cases/inference/completion.json

@@ -4,6 +4,12 @@
             "content": "Complete the sentence using one word: Roses are red, violets are "
         }
     },
+    "suffix": {
+        "data": {
+            "content": "The capital of ",
+            "suffix": "is Paris."
+        }
+    },
     "non_streaming": {
         "data": {
             "content": "Micheael Jordan is born in ",