fix: align ET budget error strings with gh-aw detection patterns (#2770)

* Initial plan

* fix: use 'exceeded' in ET limit event type and message

Update the event type, JSON error body type, and error message in
proxy-request.js to use 'exceeded' terminology instead of 'reached'.

This aligns AWF log events and error responses with the detection
patterns in gh-aw's parse_mcp_gateway_log.cjs, which require 'exceeded'
after 'effective tokens'. The previous 'reached' wording caused all
three detection regexes to fail on AWF log lines:

  Before: event=effective_tokens_limit_reached
          type=effective_tokens_limit_reached
          message='Maximum effective tokens reached (X / Y).'

  After:  event=effective_tokens_limit_exceeded
          type=effective_tokens_limit_exceeded
          message='Maximum effective tokens exceeded (X / Y).'

The updated strings satisfy the existing gh-aw pattern:
  /effective[\s_-]*tokens?.*exceeded/i

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

Author: Copilot

containers/api-proxy/proxy-request.js

@@ -263,8 +263,8 @@ function resetEffectiveTokenGuardForTests() {
 function buildEffectiveTokenLimitError(etState) {
   return {
     error: {
-      type: 'effective_tokens_limit_reached',
-      message: `Maximum effective tokens reached (${etState.totalEffectiveTokens.toFixed(2)} / ${etState.maxEffectiveTokens}).`,
+      type: 'effective_tokens_limit_exceeded',
+      message: `Maximum effective tokens exceeded (${etState.totalEffectiveTokens.toFixed(2)} / ${etState.maxEffectiveTokens}).`,
       total_effective_tokens: etState.totalEffectiveTokens,
       max_effective_tokens: etState.maxEffectiveTokens,
     },
@@ -483,7 +483,7 @@ function proxyRequest(req, res, targetHost, injectHeaders, provider, basePath =
       metrics.gaugeDec('active_requests', { provider });
       metrics.increment('requests_total', { provider, method: req.method, status_class: '4xx' });
       metrics.observe('request_duration_ms', duration, { provider });
-      logRequest('warn', 'effective_tokens_limit_reached', {
+      logRequest('warn', 'effective_tokens_limit_exceeded', {
         request_id: requestId,
         provider,
         total_effective_tokens: etBlock.totalEffectiveTokens,
@@ -628,7 +628,7 @@ function proxyWebSocket(req, socket, head, targetHost, injectHeaders, provider,
 
   const etBlock = getEffectiveTokenBlockState();
   if (etBlock && etBlock.maxExceeded) {
-    logRequest('warn', 'effective_tokens_limit_reached', {
+    logRequest('warn', 'effective_tokens_limit_exceeded', {
       request_id: requestId,
       provider,
       total_effective_tokens: etBlock.totalEffectiveTokens,

containers/api-proxy/server.proxy.test.js

@@ -495,7 +495,7 @@ describe('proxyRequest effective token guard', () => {
       'Content-Type': 'application/json',
     }));
     const payload = JSON.parse(res2.end.mock.calls[0][0]);
-    expect(payload.error.type).toBe('effective_tokens_limit_reached');
+    expect(payload.error.type).toBe('effective_tokens_limit_exceeded');
     expect(payload.error.max_effective_tokens).toBe(10);
     expect(payload.error.total_effective_tokens).toBeGreaterThanOrEqual(10);
   });