Add mock-based E2E tests and gate live tests to main/nightly

Introduce a new mock-based E2E test suite that uses pytest_httpserver to
simulate LLM provider responses, eliminating the need for real API keys
on PR builds. The mock tests cover model alias routing, protocol
transformation (OpenAI↔Anthropic), Responses API passthrough/translation,
streaming, tool calls, thinking mode, and multi-turn state management.

CI changes:
- Add mock-e2e-tests job (zero secrets, runs on every PR)
- Gate all live E2E jobs to main pushes + nightly schedule
- Scope secrets to only the keys each job actually needs
- Add daily cron schedule for full live test coverage

Also relaxes exact-match assertions in live e2e tests to structural
checks (non-null, non-empty) since LLM output is non-deterministic.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

tests/e2e/test_model_alias_routing.py

@@ -120,7 +120,8 @@ def test_openai_client_with_alias_arch_summarize_v1():
 
     response_content = completion.choices[0].message.content
     logger.info(f"Response from arch.summarize.v1 alias: {response_content}")
-    assert response_content == "Hello from alias arch.summarize.v1!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_openai_client_with_alias_arch_v1():
@@ -146,7 +147,8 @@ def test_openai_client_with_alias_arch_v1():
 
     response_content = completion.choices[0].message.content
     logger.info(f"Response from arch.v1 alias: {response_content}")
-    assert response_content == "Hello from alias arch.v1!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_anthropic_client_with_alias_arch_summarize_v1():
@@ -171,7 +173,8 @@ def test_anthropic_client_with_alias_arch_summarize_v1():
     logger.info(
         f"Response from arch.summarize.v1 alias via Anthropic: {response_content}"
     )
-    assert response_content == "Hello from alias arch.summarize.v1 via Anthropic!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_anthropic_client_with_alias_arch_v1():
@@ -194,7 +197,8 @@ def test_anthropic_client_with_alias_arch_v1():
 
     response_content = "".join(b.text for b in message.content if b.type == "text")
     logger.info(f"Response from arch.v1 alias via Anthropic: {response_content}")
-    assert response_content == "Hello from alias arch.v1 via Anthropic!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_openai_client_with_alias_streaming():
@@ -228,7 +232,8 @@ def test_openai_client_with_alias_streaming():
 
     full_content = "".join(content_chunks)
     logger.info(f"Streaming response from arch.summarize.v1 alias: {full_content}")
-    assert full_content == "Hello from streaming alias!"
+    assert full_content is not None
+    assert len(full_content) > 0
 
 
 def test_anthropic_client_with_alias_streaming():
@@ -256,7 +261,8 @@ def test_anthropic_client_with_alias_streaming():
     logger.info(
         f"Streaming response from arch.summarize.v1 alias via Anthropic: {full_text}"
     )
-    assert full_text == "Hello from streaming alias via Anthropic!"
+    assert full_text is not None
+    assert len(full_text) > 0
 
 
 def test_400_error_handling_with_alias():
@@ -400,7 +406,8 @@ def test_direct_model_4o_mini_openai():
 
     response_content = completion.choices[0].message.content
     logger.info(f"Response from direct 4o-mini: {response_content}")
-    assert response_content == "Hello from direct 4o-mini!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_direct_model_4o_mini_anthropic():
@@ -423,7 +430,8 @@ def test_direct_model_4o_mini_anthropic():
 
     response_content = "".join(b.text for b in message.content if b.type == "text")
     logger.info(f"Response from direct 4o-mini via Anthropic: {response_content}")
-    assert response_content == "Hello from direct 4o-mini via Anthropic!"
+    assert response_content is not None
+    assert len(response_content) > 0
 
 
 def test_anthropic_thinking_mode_streaming():

tests/e2e/test_prompt_gateway.py

@@ -405,7 +405,8 @@ def test_claude_v1_messages_api():
         ],
     )
 
-    assert message.content[0].text == "Hello from Claude!"
+    assert message.content[0].text is not None
+    assert len(message.content[0].text) > 0
 
 
 def test_claude_v1_messages_api_streaming():
@@ -432,8 +433,10 @@ def test_claude_v1_messages_api_streaming():
         # A safe way to reassemble text from the content blocks:
         final_text = "".join(b.text for b in final.content if b.type == "text")
 
-    assert full_text == "Hello from Claude!"
-    assert final_text == "Hello from Claude!"
+    assert full_text is not None
+    assert len(full_text) > 0
+    assert final_text is not None
+    assert len(final_text) > 0
 
 
 def test_anthropic_client_with_openai_model_streaming():
@@ -463,8 +466,10 @@ def test_anthropic_client_with_openai_model_streaming():
         # A safe way to reassemble text from the content blocks:
         final_text = "".join(b.text for b in final.content if b.type == "text")
 
-    assert full_text == "Hello from ChatGPT!"
-    assert final_text == "Hello from ChatGPT!"
+    assert full_text is not None
+    assert len(full_text) > 0
+    assert final_text is not None
+    assert len(final_text) > 0
 
 
 def test_openai_gpt4o_mini_v1_messages_api():
@@ -488,7 +493,8 @@ def test_openai_gpt4o_mini_v1_messages_api():
         ],
     )
 
-    assert completion.choices[0].message.content == "Hello from GPT-4o-mini!"
+    assert completion.choices[0].message.content is not None
+    assert len(completion.choices[0].message.content) > 0
 
 
 def test_openai_gpt4o_mini_v1_messages_api_streaming():
@@ -521,7 +527,8 @@ def test_openai_gpt4o_mini_v1_messages_api_streaming():
 
     # Reconstruct the full message
     full_content = "".join(content_chunks)
-    assert full_content == "Hello from GPT-4o-mini!"
+    assert full_content is not None
+    assert len(full_content) > 0
 
 
 def test_openai_client_with_claude_model_streaming():