Add PerplexityLLMAdapter to enforce Perplexity's message ordering constraints

Perplexity's API is stricter than OpenAI about conversation history:
- Requires strict alternation between user/tool and assistant messages
- Disallows system messages except as the initial message
- Requires the last message to be user or tool

The new adapter transforms messages before sending to satisfy all three
constraints: merging consecutive initial system messages, converting
non-initial system to user, merging consecutive same-role messages, and
removing trailing assistant messages.

Also adds dual-system-instruction warnings to Cerebras, Fireworks,
Mistral, Perplexity, and SambaNova services (matching the existing
BaseOpenAILLMService pattern), and updates the warning text in
BaseOpenAILLMService to be more descriptive.

src/pipecat/services/cerebras/llm.py

@@ -117,6 +117,10 @@ class CerebrasLLMService(OpenAILLMService):
         # Prepend system instruction if set
         if self._settings.system_instruction:
             messages = params.get("messages", [])
+            if messages and messages[0].get("role") == "system":
+                logger.warning(
+                    f"{self}: Both system_instruction and an initial system message in context are set. This may be unintended."
+                )
             params["messages"] = [
                 {"role": "system", "content": self._settings.system_instruction}
             ] + messages