fix(security): consolidated security hardening — SSRF, timing attack, tar traversal, credential leakage (#5944)

Salvaged from PRs #5800 (memosr), #5806 (memosr), #5915 (Ruzzgar), #5928 (Awsh1).

Changes:
- Use hmac.compare_digest for API key comparison (timing attack prevention)
- Apply provider env var blocklist to Docker containers (credential leakage)
- Replace tar.extractall() with safe extraction in TerminalBench2 (CVE-2007-4559)
- Add SSRF protection via is_safe_url to ALL platform adapters:
  base.py (cache_image_from_url, cache_audio_from_url),
  discord, slack, telegram, matrix, mattermost, feishu, wecom
  (Signal and WhatsApp protected via base.py helpers)
- Update tests: mock is_safe_url in Mattermost download tests
- Add security tests for tar extraction (traversal, symlinks, safe files)

gateway/platforms/discord.py

@@ -55,6 +55,7 @@ from gateway.platforms.base import (
     cache_document_from_bytes,
     SUPPORTED_DOCUMENT_TYPES,
 )
+from tools.url_safety import is_safe_url
 
 
 def _clean_discord_id(entry: str) -> str:
@@ -1285,6 +1286,10 @@ class DiscordAdapter(BasePlatformAdapter):
         if not self._client:
             return SendResult(success=False, error="Not connected")
 
+        if not is_safe_url(image_url):
+            logger.warning("[%s] Blocked unsafe image URL during Discord send_image", self.name)
+            return await super().send_image(chat_id, image_url, caption, reply_to, metadata=metadata)
+
         try:
             import aiohttp