feat(antigravity): comprehensive enhancements - model mapping, rate limiting, scheduling & ops

Key changes:
- Upgrade model mapping: Opus 4.5 → Opus 4.6-thinking with precise matching
- Unified rate limiting: scope-level → model-level with Redis snapshot sync
- Load-balanced scheduling by call count with smart retry mechanism
- Force cache billing support
- Model identity injection in prompts with leak prevention
- Thinking mode auto-handling (max_tokens/budget_tokens fix)
- Frontend: whitelist mode toggle, model mapping validation, status indicators
- Gemini session fallback with Redis Trie O(L) matching
- Ops: enhanced concurrency monitoring, account availability, retry logic
- Migration scripts: 049-051 for model mapping unification

backend/internal/service/ops_service_redaction_test.go

@@ -0,0 +1,99 @@
+package service
+
+import (
+	"encoding/json"
+	"testing"
+)
+
+func TestIsSensitiveKey_TokenBudgetKeysNotRedacted(t *testing.T) {
+	t.Parallel()
+
+	for _, key := range []string{
+		"max_tokens",
+		"max_output_tokens",
+		"max_input_tokens",
+		"max_completion_tokens",
+		"max_tokens_to_sample",
+		"budget_tokens",
+		"prompt_tokens",
+		"completion_tokens",
+		"input_tokens",
+		"output_tokens",
+		"total_tokens",
+		"token_count",
+	} {
+		if isSensitiveKey(key) {
+			t.Fatalf("expected key %q to NOT be treated as sensitive", key)
+		}
+	}
+
+	for _, key := range []string{
+		"authorization",
+		"Authorization",
+		"access_token",
+		"refresh_token",
+		"id_token",
+		"session_token",
+		"token",
+		"client_secret",
+		"private_key",
+		"signature",
+	} {
+		if !isSensitiveKey(key) {
+			t.Fatalf("expected key %q to be treated as sensitive", key)
+		}
+	}
+}
+
+func TestSanitizeAndTrimRequestBody_PreservesTokenBudgetFields(t *testing.T) {
+	t.Parallel()
+
+	raw := []byte(`{"model":"claude-3","max_tokens":123,"thinking":{"type":"enabled","budget_tokens":456},"access_token":"abc","messages":[{"role":"user","content":"hi"}]}`)
+	out, _, _ := sanitizeAndTrimRequestBody(raw, 10*1024)
+	if out == "" {
+		t.Fatalf("expected non-empty sanitized output")
+	}
+
+	var decoded map[string]any
+	if err := json.Unmarshal([]byte(out), &decoded); err != nil {
+		t.Fatalf("unmarshal sanitized output: %v", err)
+	}
+
+	if got, ok := decoded["max_tokens"].(float64); !ok || got != 123 {
+		t.Fatalf("expected max_tokens=123, got %#v", decoded["max_tokens"])
+	}
+
+	thinking, ok := decoded["thinking"].(map[string]any)
+	if !ok || thinking == nil {
+		t.Fatalf("expected thinking object to be preserved, got %#v", decoded["thinking"])
+	}
+	if got, ok := thinking["budget_tokens"].(float64); !ok || got != 456 {
+		t.Fatalf("expected thinking.budget_tokens=456, got %#v", thinking["budget_tokens"])
+	}
+
+	if got := decoded["access_token"]; got != "[REDACTED]" {
+		t.Fatalf("expected access_token to be redacted, got %#v", got)
+	}
+}
+
+func TestShrinkToEssentials_IncludesThinking(t *testing.T) {
+	t.Parallel()
+
+	root := map[string]any{
+		"model":      "claude-3",
+		"max_tokens": 100,
+		"thinking": map[string]any{
+			"type":          "enabled",
+			"budget_tokens": 200,
+		},
+		"messages": []any{
+			map[string]any{"role": "user", "content": "first"},
+			map[string]any{"role": "user", "content": "last"},
+		},
+	}
+
+	out := shrinkToEssentials(root)
+	if _, ok := out["thinking"]; !ok {
+		t.Fatalf("expected thinking to be included in essentials: %#v", out)
+	}
+}