refactor(acms): unify ContextFragment model hierarchies by extending CRP base types

Core domain types (FragmentProvenance, ContextFragment, ContextBudget,
ContextPayload) now extend their CRP counterparts via Pydantic v2
inheritance, ensuring isinstance compatibility across the model
hierarchy.

Key changes:
- CRP base types made frozen=True (no consumer mutates them)
- CRP AssembledContext fields changed from list to tuple (frozen consistency)
- Core types extend CRP bases: FragmentProvenance(CRPFragmentProvenance),
  ContextFragment(CRPContextFragment), ContextBudget(CRPContextBudget),
  ContextPayload(CRPAssembledContext)
- Removed duplicate ContextFragment dataclass from skeleton_compressor
- Updated project_context.py to pass tuples to frozen AssembledContext
- Added Behave tests (10 scenarios), Robot integration tests (3 cases),
  and ASV benchmarks for the unified hierarchy
- Updated Known Limitations table in docs/reference/acms.md

ISSUES CLOSED: #569

benchmarks/skeleton_compressor_bench.py

@@ -22,9 +22,15 @@ import cleveragents  # noqa: E402
 importlib.reload(cleveragents)
 
 from cleveragents.application.services.skeleton_compressor import (  # noqa: E402
-    ContextFragment,
     SkeletonCompressorService,
 )
+from cleveragents.domain.models.core.context_fragment import (  # noqa: E402
+    ContextFragment,
+    FragmentProvenance,
+)
+
+# Default provenance for skeleton benchmark fragments.
+_SKEL_PROV = FragmentProvenance(resource_uri="bench-skeleton://default")
 
 
 def _build_fragments(count: int, tokens_each: int = 100) -> list[ContextFragment]:
@@ -32,10 +38,14 @@ def _build_fragments(count: int, tokens_each: int = 100) -> list[ContextFragment
     return [
         ContextFragment(
             fragment_id=f"bench-{i:05d}",
+            uko_node=f"bench-skeleton://file/{i}",
             content="x" * tokens_each,
             token_count=tokens_each,
-            relevance=round(1.0 - (i / max(count, 1)), 4),
-            source_decision_id=f"01HX{'B' * 22}{i:02d}" if i < 99 else None,
+            relevance_score=round(1.0 - (i / max(count, 1)), 4),
+            provenance=_SKEL_PROV,
+            metadata=(
+                {"source_decision_id": f"01HX{'B' * 22}{i:02d}"} if i < 99 else {}
+            ),
         )
         for i in range(count)
     ]