feat(gateway): harden failover and payload handling

Retry pre-response transport failures across candidates with an explicit stop policy, and propagate end-to-end timing into usage records and UI diagnostics.

Remove legacy body, import, cookie, PII, and tunnel replay caps while preserving optional operator-configured gateway limits.

frontend/src/i18n/__tests__/i18n.spec.ts

@@ -114,6 +114,7 @@ describe('i18n infrastructure', () => {
     expect(translateLegacyText('请求记录清理策略', 'en-US')).toBe('Request log cleanup policy')
     expect(translateLegacyText('最大转移次数', 'en-US')).toBe('Max transfers')
     expect(translateLegacyText('最大转移超时', 'en-US')).toBe('Max transfer timeout')
+    expect(translateLegacyText('0 (不限制)', 'en-US')).toBe('0 (unlimited)')
     expect(translateLegacyText('  发布于 2026-01-01  ', 'en-US')).toBe('  Published at 2026-01-01  ')
     expect(translateLegacyText('git clone https://github.com/fawney19/Aether.git', 'en-US')).toBe('git clone https://github.com/fawney19/Aether.git')
   })

frontend/src/i18n/messages.ts

@@ -1412,6 +1412,7 @@ const legacyExactEnglishMessages: Record<string, string> = {
   '默认 2': 'Default 2',
   '最大转移次数': 'Max transfers',
   '最大转移超时': 'Max transfer timeout',
+  '0 (不限制)': '0 (unlimited)',
   '流式首字节超时': 'Streaming first-byte timeout',
   '非流式请求超时': 'Non-streaming request timeout',
   '(秒)': '(seconds)',
@@ -2409,12 +2410,8 @@ const legacyPhraseEnglishMessages: Array<[string, string]> = [
   ['记录详细程度', 'Record detail level'],
   ['BASIC - 基本信息 (~1KB/条)', 'BASIC - basic information (~1KB/item)'],
   ['HEADERS - 含请求头 (~2-3KB/条)', 'HEADERS - includes request headers (~2-3KB/item)'],
-  ['FULL - 完整请求响应 (~50KB/条)', 'FULL - full request and response (~50KB/item)'],
+  ['FULL - 完整请求响应', 'FULL - full request and response'],
   ['敏感信息会自动脱敏', 'Sensitive information is automatically redacted'],
-  ['最大请求体大小 (KB)', 'Maximum request body size (KB)'],
-  ['超过此大小的请求体将被截断记录', 'Request bodies larger than this are truncated in records'],
-  ['最大响应体大小 (KB)', 'Maximum response body size (KB)'],
-  ['超过此大小的响应体将被截断记录', 'Response bodies larger than this are truncated in records'],
   ['敏感请求头', 'Sensitive request headers'],
   ['逗号分隔，这些请求头会被脱敏处理', 'Comma-separated request headers that will be redacted'],
   ['0 表示默认不限制；未单独配置的用户和独立 Key 会跟随这里', '0 means no default limit. Users and standalone keys without separate configuration follow this setting.'],
@@ -2750,10 +2747,6 @@ const legacyFallbackTokens: Array<[string, string]> = [
   ['完整请求响应', 'full request and response'],
   ['自动脱敏', 'automatically redacted'],
   ['脱敏', 'redacted'],
-  ['最大请求体', 'maximum request body'],
-  ['最大响应体', 'maximum response body'],
-  ['超过此大小', 'larger than this'],
-  ['截断记录', 'truncated in records'],
   ['逗号分隔', 'comma-separated'],
   ['这些请求头', 'these request headers'],
   ['脱敏处理', 'redacted'],