Update the "InterleavedTextMedia" type (#635)

## What does this PR do?

This is a long-pending change and particularly important to get done
now.

Specifically:
- we cannot "localize" (aka download) any URLs from media attachments
anywhere near our modeling code. it must be done within llama-stack.
- `PIL.Image` is infesting all our APIs via `ImageMedia ->
InterleavedTextMedia` and that cannot be right at all. Anything in the
API surface must be "naturally serializable". We need a standard `{
type: "image", image_url: "<...>" }` which is more extensible
- `UserMessage`, `SystemMessage`, etc. are moved completely to
llama-stack from the llama-models repository.

See https://github.com/meta-llama/llama-models/pull/244 for the
corresponding PR in llama-models.

## Test Plan

```bash
cd llama_stack/providers/tests

pytest -s -v -k "fireworks or ollama or together" inference/test_vision_inference.py
pytest -s -v -k "(fireworks or ollama or together) and llama_3b" inference/test_text_inference.py
pytest -s -v -k chroma memory/test_memory.py \
  --env EMBEDDING_DIMENSION=384 --env CHROMA_DB_PATH=/tmp/foobar

pytest -s -v -k fireworks agents/test_agents.py  \
   --safety-shield=meta-llama/Llama-Guard-3-8B \
   --inference-model=meta-llama/Llama-3.1-8B-Instruct
```

Updated the client sdk (see PR ...), installed the SDK in the same
environment and then ran the SDK tests:

```bash
cd tests/client-sdk
LLAMA_STACK_CONFIG=together pytest -s -v agents/test_agents.py
LLAMA_STACK_CONFIG=ollama pytest -s -v memory/test_memory.py

# this one needed a bit of hacking in the run.yaml to ensure I could register the vision model correctly
INFERENCE_MODEL=llama3.2-vision:latest LLAMA_STACK_CONFIG=ollama pytest -s -v inference/test_inference.py
```

llama_stack/providers/inline/safety/code_scanner/code_scanner.py

@@ -7,13 +7,17 @@
 import logging
 from typing import Any, Dict, List
 
-from llama_models.llama3.api.datatypes import interleaved_text_media_as_str, Message
+from llama_stack.apis.safety import *  # noqa: F403
+from llama_stack.apis.inference import Message
+from llama_stack.providers.utils.inference.prompt_adapter import (
+    interleaved_content_as_str,
+)
 
 from .config import CodeScannerConfig
 
-from llama_stack.apis.safety import *  # noqa: F403
 
 log = logging.getLogger(__name__)
+
 ALLOWED_CODE_SCANNER_MODEL_IDS = [
     "CodeScanner",
     "CodeShield",
@@ -48,7 +52,7 @@ class MetaReferenceCodeScannerSafetyImpl(Safety):
 
         from codeshield.cs import CodeShield
 
-        text = "\n".join([interleaved_text_media_as_str(m.content) for m in messages])
+        text = "\n".join([interleaved_content_as_str(m.content) for m in messages])
         log.info(f"Running CodeScannerShield on {text[50:]}")
         result = await CodeShield.scan_code(text)
 

llama_stack/providers/inline/safety/llama_guard/llama_guard.py

@@ -12,9 +12,13 @@ from typing import Any, Dict, List, Optional
 from llama_models.llama3.api.datatypes import *  # noqa: F403
 from llama_stack.apis.inference import *  # noqa: F403
 from llama_stack.apis.safety import *  # noqa: F403
+from llama_stack.apis.common.content_types import ImageContentItem, TextContentItem
 from llama_stack.distribution.datatypes import Api
 
 from llama_stack.providers.datatypes import ShieldsProtocolPrivate
+from llama_stack.providers.utils.inference.prompt_adapter import (
+    interleaved_content_as_str,
+)
 
 from .config import LlamaGuardConfig
 
@@ -258,18 +262,18 @@ class LlamaGuardShield:
         most_recent_img = None
 
         for m in messages[::-1]:
-            if isinstance(m.content, str):
+            if isinstance(m.content, str) or isinstance(m.content, TextContentItem):
                 conversation.append(m)
-            elif isinstance(m.content, ImageMedia):
+            elif isinstance(m.content, ImageContentItem):
                 if most_recent_img is None and m.role == Role.user.value:
                     most_recent_img = m.content
                     conversation.append(m)
             elif isinstance(m.content, list):
                 content = []
                 for c in m.content:
-                    if isinstance(c, str):
+                    if isinstance(c, str) or isinstance(c, TextContentItem):
                         content.append(c)
-                    elif isinstance(c, ImageMedia):
+                    elif isinstance(c, ImageContentItem):
                         if most_recent_img is None and m.role == Role.user.value:
                             most_recent_img = c
                             content.append(c)
@@ -292,7 +296,7 @@ class LlamaGuardShield:
         categories_str = "\n".join(categories)
         conversations_str = "\n\n".join(
             [
-                f"{m.role.capitalize()}: {interleaved_text_media_as_str(m.content)}"
+                f"{m.role.capitalize()}: {interleaved_content_as_str(m.content)}"
                 for m in messages
             ]
         )

llama_stack/providers/inline/safety/prompt_guard/prompt_guard.py

@@ -17,6 +17,9 @@ from llama_stack.apis.safety import *  # noqa: F403
 from llama_models.llama3.api.datatypes import *  # noqa: F403
 
 from llama_stack.providers.datatypes import ShieldsProtocolPrivate
+from llama_stack.providers.utils.inference.prompt_adapter import (
+    interleaved_content_as_str,
+)
 
 from .config import PromptGuardConfig, PromptGuardType
 
@@ -83,7 +86,7 @@ class PromptGuardShield:
 
     async def run(self, messages: List[Message]) -> RunShieldResponse:
         message = messages[-1]
-        text = interleaved_text_media_as_str(message.content)
+        text = interleaved_content_as_str(message.content)
 
         # run model on messages and return response
         inputs = self.tokenizer(text, return_tensors="pt")