fix(glm): default effort xhigh, document real GLM 5.3 effort mapping

Z.ai Anthropic endpoint reads output_config.effort and accepts all 5
Claude Code levels, but GLM 5.3 has only low/high/max natively:
medium -> low, xhigh -> high (measured, not documented by Z.ai).
ai-glm now starts on xhigh instead of max.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/test_fixes.sh

@@ -104,6 +104,16 @@ test_glm_official_api() {
   fi
 }
 
+# ── ai-glm: default effort is xhigh (still overridable via AI_EFFORT) ─────
+test_glm_default_effort() {
+  if echo "$GLM_SECTION" | grep -q -- '--effort "${AI_EFFORT:-xhigh}"' \
+    && echo "$GLM_SECTION" | grep -q 'CLAUDE_CODE_ALWAYS_ENABLE_EFFORT=1'; then
+    ok "ai-glm: starts with --effort xhigh by default, effort sent to Z.ai"
+  else
+    fail "ai-glm: must start with --effort \"\${AI_EFFORT:-xhigh}\" and CLAUDE_CODE_ALWAYS_ENABLE_EFFORT=1"
+  fi
+}
+
 # ── ai-gemini: native launcher generation ─────────────────────────────────
 test_gemini_native_launcher() {
   if echo "$GEMINI_SECTION" | grep -q 'antigravity CLI (agy)' \
@@ -301,6 +311,7 @@ test_kimi_official_api
 test_kimi_no_artemox
 test_glm_claude_launcher
 test_glm_official_api
+test_glm_default_effort
 test_gemini_native_launcher
 test_cursor_native_launcher
 test_global_rules_source_installed