feat: introduces session-centric tracing APIs (inclusionAI#539)

* Introduces session-centric tracing APIs

Replaces request tracing with a session-scoped tracer so rollout executors, engines, and workflows can log lifecycle and phase spans in a single record. Adds context-aware helpers and decorators to propagate session IDs through async code, enabling phase timing, counter aggregation, and structured flush thresholds. Updates configs, tests, docs, and example presets to align with the session tracer and tightens validation around workflow submission and tracing hooks.

Adds request phase tracing helpers (inclusionAI#543)

* Resolve review comments

Author: rchardx

areal/api/cli_args.py

@@ -1001,23 +1001,23 @@ class StatsLoggerConfig:
 
 
 @dataclass
-class RequestTracerConfig:
-    """Configuration for per-request lifecycle tracing."""
+class SessionTracerConfig:
+    """Configuration for per-session lifecycle tracing."""
 
     enabled: bool = field(
         default=False,
         metadata={
             "help": (
-                "Enable per-request lifecycle tracing alongside perf events. "
-                "When true, request metadata is captured to requests.jsonl."
+                "Enable per-session lifecycle tracing alongside perf events. "
+                "When true, session metadata is captured to sessions.jsonl."
             )
         },
     )
     flush_threshold: int = field(
         default=256,
         metadata={
             "help": (
-                "Flush request trace records once this many entries are ready. "
+                "Flush session trace records once this many entries are ready. "
                 "Values <= 0 fall back to 1."
             )
         },
@@ -1048,9 +1048,9 @@ class PerfTracerConfig:
             )
         },
     )
-    request_tracer: RequestTracerConfig | None = field(
+    session_tracer: SessionTracerConfig | None = field(
         default=None,
-        metadata={"help": "Request tracing configuration."},
+        metadata={"help": "Session tracing configuration."},
     )
 
 

areal/api/io_struct.py

@@ -30,7 +30,7 @@ class ModelRequest:
     tokenizer: PreTrainedTokenizerFast | None = None
 
     # vlm
-    image_data: list[ImageObject | str] | None = field(default_factory=list)
+    image_data: list[str] | None = field(default_factory=list)
     processor: Optional["AutoProcessor"] = None
 
     def copy(self):

areal/api/workflow_api.py

@@ -1,5 +1,6 @@
 from __future__ import annotations  # noqa
 
+from abc import ABC, abstractmethod
 from typing import TYPE_CHECKING, Any
 
 from areal.experimental.openai.types import InteractionWithTokenLogpReward
@@ -8,7 +9,8 @@
     from areal.api.engine_api import InferenceEngine
 
 
-class RolloutWorkflow:
+class RolloutWorkflow(ABC):
+    @abstractmethod
     async def arun_episode(
         self, engine: InferenceEngine, data: dict[str, Any]
     ) -> dict[str, Any] | None | dict[str, InteractionWithTokenLogpReward]: