What does this PR do?

gateway/platforms/weixin.py passed media URLs from the WeChat API
response directly to _download_bytes() without any validation:

# Before (vulnerable)
elif full_url:
    raw = await _download_bytes(session, url=full_url, ...)

full_url comes directly from the WeChat API JSON response. If the
API response is manipulated (MitM, compromised CDN, malicious bot
configuration), an attacker could redirect downloads to:

• http://169.254.169.254/latest/meta-data/ → AWS IMDS
• http://192.168.x.x/admin → internal network services
• file:///etc/passwd → local file read

Fix

Added _WEIXIN_CDN_ALLOWLIST — a frozenset of known WeChat CDN
hostnames — and _assert_weixin_cdn_url() that validates every
media URL before download:

_WEIXIN_CDN_ALLOWLIST = frozenset({
    "novac2c.cdn.weixin.qq.com",
    "ilinkai.weixin.qq.com",
    "wx.qlogo.cn",
    "thirdwx.qlogo.cn",
    "res.wx.qq.com",
    "mmbiz.qpic.cn",
    "mmbiz.qlogo.cn",
})

Rejects non-http(s) schemes and hosts outside the allowlist with
a ValueError before any network request is made.

Type of Change

• 🔒 Security fix (SSRF)

Checklist

• Read the Contributing Guide
• Commit messages follow Conventional Commits
• Consistent with existing SSRF protection patterns in Hermes
• No behavior change for legitimate WeChat CDN URLs