fix: replace RuntimeError with graceful degradation for unverified LM Studio loads

Salvage of PR #52188. The original PR raised RuntimeError when LM Studio
load was rejected or unverifiable, which would abort agent startup on
transient network failures. Replace with logger.warning + fallback to
configured context length, preserving the old graceful-degradation behavior.

agent/agent_init.py

@@ -2280,9 +2280,9 @@ def init_agent(
         _config_context_length
     )
     if agent._lmstudio_load_was_unverified(_lmstudio_runtime_context_length):
-        raise RuntimeError(
+        _ra().logger.warning(
             "LM Studio model activation was rejected or completed without a "
-            "verifiable active context length; agent startup aborted"
+            "verifiable active context length; falling back to configured context"
         )
     _effective_context_length = agent._effective_lmstudio_context_length(
         _config_context_length,

agent/agent_runtime_helpers.py

@@ -2341,10 +2341,10 @@ def switch_model(agent, new_model, new_provider, api_key='', base_url='', api_mo
         _destination_context_intent
     )
     if agent._lmstudio_load_was_unverified(_runtime_context_length):
-        _restore_snapshot()
-        raise RuntimeError(
+        logger.warning(
             "LM Studio model activation was rejected or completed without a "
-            "verifiable active context length; model switch aborted"
+            "verifiable active context length during model switch; continuing "
+            "with configured context"
         )
     _effective_context_length = agent._effective_lmstudio_context_length(
         _destination_context_intent,