docs: measure the 1.3.3 notes against 1.3.2, not against this branch

The behaviour section described round trips that only ever existed inside the
branch. Verified every claim against `git show 6ec9ec7:<file>` plus a 300-combo
side-by-side run (4 old provider slots x 25 models x three ENABLE_THINKING
states); exactly two differences exist.

- glm-5.3 / glm-5.3-flash / gemini-3.1-pro are newly registered as unable to
  stop reasoning. They were unregistered in 1.3.2, so ENABLE_THINKING=false
  passed; now it fails at assembly. That is the one change that breaks an
  existing config, so it leads the table.
- The openai slot moves from "shape unknown, refuse at assembly" to the standard
  OpenAI shapes in both directions. Neither openai nor anthropic nor google ever
  injected medium; minimax did and still does, byte for byte.
- kimi-k3 was never registered before, so it changes nothing for anyone. Moved
  to new capabilities as a first registration.
- Both keyword and positional ThinkingCapability construction break, not just
  positional; spell out the migration form.

.env.example: name all eight registry slots, and say that "on" injects nothing
on the three slots whose on_base is empty.

.env.example

@@ -2,7 +2,8 @@
 # 键名清单 = M1 设计文档 §8 定稿;缺关键配置直接报错,不做默认值兜底。
 
 # ══ 多源配置: {SCOPE}__{PROVIDER}__{N}__{FIELD} ══
-# PROVIDER 必须是注册表键(qwen/deepseek/openai,或 register_provider 注册后经 registry 传入)。
+# PROVIDER 必须是注册表键(八段: qwen/deepseek/zhipu/moonshot/minimax/openai/anthropic/google,
+#   或 register_provider 注册后经 registry 传入)。
 # 必填: BASE_URL / API_KEY / MODEL / TIMEOUT_S(或用平铺 LLM_TIMEOUT 作缺省)。
 LLM__QWEN__1__BASE_URL=
 LLM__QWEN__1__API_KEY=
@@ -15,8 +16,11 @@ LLM__QWEN__1__TIMEOUT_S=120
 # LLM__QWEN__1__EST_TOKENS=2000          # 可选调优覆盖: TPM 入场预扣量;未填则库按 tpm//60 派生
 # LLM__QWEN__1__TTFT_TIMEOUT_S=30        # 须与 INTER_TOKEN 成对;0 < inter < ttft < timeout
 # LLM__QWEN__1__INTER_TOKEN_TIMEOUT_S=15
-# LLM__QWEN__1__ENABLE_THINKING=true     # 三态: 缺省=不注入 / true=注入开启 / false=注入关闭
+# LLM__QWEN__1__ENABLE_THINKING=true     # 三态: 缺省=不表态 / true=要求开启 / false=要求关闭
 #   本键是 REASONING_EFFORT 的语法糖: true ≡ auto、false ≡ none、缺省 ≡ 不表态
+#   "要求开启"注入什么随 provider 段而定: openai/anthropic/google 三段的开启形态是
+#   on_base={}——一个字节都不注入,走模型自己的默认档(该默认档若不推理,本键不会报错
+#   也不会开推理,见 CHANGELOG 1.3.3「已知限制」/ issue #21);要确保开启请配 REASONING_EFFORT
 # LLM__QWEN__1__REASONING_EFFORT=low     # 本源默认推理档位;缺省=不表态(随模型自己的默认档)
 #   八档(封闭词汇): none | auto | minimal | low | medium | high | xhigh | max
 #     none = 要求不推理(与"缺省不表态"是两回事);auto = 要求推理但不指定强度