wechatapp.py has no allowed_users gating, unlike every other adapter

[shaun0927]

Observation

Every other chat-app adapter in frontends/ reads an allow-list from
mykey.py and either rejects or logs unknown users:

Adapter	Reads allow-list	Behavior on empty list
tgapp.py	tg_allowed_users	exit() (line 122-124)
fsapp.py	fs_allowed_users	public if [*]
qqapp.py	qq_allowed_users	public if [*]
dingtalkapp.py	dingtalk_allowed_users	public if [*]
wecomapp.py	wecom_allowed_users	public if [*]
wechatapp.py	(none — keyword grep returns 0 matches)	n/a — no gate at all

README.md (Chinese half, line 297) tells the user the wechat adapter
needs no extra config: "扫码登录即可". Today that means after the QR
login, any contact who messages the bound WeChat account drives the
agent — code_run, file_read, file_write, etc. all reachable.

This looks like a documentation/implementation gap rather than a design
choice, since the same author wrote the Telegram adapter that hard-exits
on an empty allow-list.

Proposal (would PR after maintainer alignment)

1. Add wechat_allowed_users = [] example to mykey_template.py with
   a short comment matching the QQ/DingTalk pattern.
2. In wechatapp.py message dispatch, drop messages whose sender openid
   isn't in the allow-list. Mirror the Telegram pattern (exit() on
   empty list) since this adapter has no "public webhook" use case
   — every user is talking to the maintainer's personal WeChat.

Two questions before I send a PR:

• Do you want the same [\"*\"] opt-in escape hatch the other adapters
  have, or follow Telegram's stricter "non-empty required" rule?
• WeChat senders are identified by openid (or the wxid surface that
  the iLink bot exposes) — should the key match the rest of the
  ecosystem (wechat_allowed_users = [\"<openid>\", ...])?

Happy to align before opening a PR so the change matches your existing
adapter convention.