fix(llm): forward reasoning effort to GPT-6 and gate minimal thinking

- reasoning_effort was forwarded only to IDs matching gpt-5 or the o-series, so
  GPT-6 models silently dropped the configured effort; match GPT-5 and later,
  with a version boundary so unrelated IDs do not match
- Gemini Pro, 3.8+ and the -latest aliases reject thinking_level "minimal"
  with a 400; send it only to numbered Flash models before 3.8 and map it to
  "low" elsewhere, since aliases move between generations

tests/test_google_thinking_level.py

@@ -1,8 +1,8 @@
 """Gemini thinking_level forwarding (Gemini 3.x).
 
 The catalog is Gemini 3.x only, which takes the string ``thinking_level``
-directly. Pro accepts low/high; Flash also accepts minimal/medium — an
-unsupported "minimal" on Pro is mapped to "low".
+directly. Pro, Gemini 3.8+ and the -latest aliases reject "minimal" with a 400,
+so it is mapped to "low" there; numbered Flash models before 3.8 accept it.
 """
 
 from unittest import mock
@@ -35,6 +35,22 @@ def test_pro_remaps_minimal_to_low():
     assert kw["thinking_level"] == "low"  # Pro doesn't accept "minimal"
 
 
+def test_flash_38_remaps_minimal_to_low():
+    kw = _captured_kwargs("gemini-3.8-flash", thinking_level="minimal")
+    assert kw["thinking_level"] == "low"  # 3.8 Flash 400s on "minimal"
+
+
+def test_flash_38_keeps_supported_levels():
+    kw = _captured_kwargs("gemini-3.8-flash", thinking_level="high")
+    assert kw["thinking_level"] == "high"
+
+
+@pytest.mark.parametrize("alias", ["gemini-flash-latest", "gemini-pro-latest"])
+def test_latest_alias_remaps_minimal_to_low(alias):
+    # Aliases move between generations; gemini-flash-latest 400s on "minimal".
+    assert _captured_kwargs(alias, thinking_level="minimal")["thinking_level"] == "low"
+
+
 def test_pro_keeps_high():
     kw = _captured_kwargs("gemini-3.1-pro-preview", thinking_level="high")
     assert kw["thinking_level"] == "high"

tests/test_openai_reasoning_effort.py

@@ -17,9 +17,10 @@ from tradingagents.llm_clients.openai_client import (
     "model,expected",
     [
         ("gpt-5.5", True), ("gpt-5.4", True), ("gpt-5.4-mini", True),
-        ("gpt-5.5-pro", True), ("o1", True), ("o3-mini", True),
+        ("gpt-5.5-pro", True), ("gpt-6-astra", True), ("o1", True), ("o3-mini", True),
         ("gpt-4.1", False), ("gpt-4o", False), ("gpt-4o-mini", False),
-        ("gpt-3.5-turbo", False),
+        ("gpt-3.5-turbo", False), ("gpt-10", True),
+        ("gpt-5foo", False), ("gpt-60x", False), ("o3rd-party", False),
     ],
 )
 def test_supports_reasoning_effort(model, expected):