Meteor 3.0 potential bug

As reported in Slack by @StorytellerCZ :

```bash

W20240122-18:59:31.199(1)? (STDERR) meteor://💻app/packages/mongo.js:3584
W20240122-18:59:31.200(1)? (STDERR) throw new Error("".concat(m, " + is not available on the server. Please use ").concat(getAsyncMethodName(m), "() instead."));
W20240122-18:59:31.200(1)? (STDERR) ^
W20240122-18:59:31.200(1)? (STDERR)
W20240122-18:59:31.200(1)? (STDERR) Error: update + is not available on the server. Please use updateAsync() instead.
W20240122-18:59:31.200(1)? (STDERR) at Object.ret. [as update] (packages/mongo/remote_collection_driver.js:52:15)
W20240122-18:59:31.200(1)? (STDERR) at Collection.update (packages/mongo/collection.js:959:31)
W20240122-18:59:31.200(1)? (STDERR) at Collection.Mongo.Collection. [as update] (packages/aldeed:collection2/main.js:236:23)
W20240122-18:59:31.200(1)? (STDERR) at packages/accounts-oauth/oauth_server.js:94:41
W20240122-18:59:31.200(1)? (STDERR) at AsynchronousCursor.forEach (packages/mongo/mongo_driver.js:1115:22)
W20240122-18:59:31.200(1)? (STDERR) at processTicksAndRejections (node:internal/process/task_queues:95:5)
W20240122-18:59:31.200(1)? (STDERR)
W20240122-18:59:31.200(1)? (STDERR) Node.js v20.9.0

```

If I add

```js
if (Meteor.isServer && Meteor.isFibersDisabled) {
return null
}
```
to line 236 in `main.js` it goes away. Probably not the best approach though.

Reproduce the Meteor 3.0 server failure using the stack trace, then inspect main.js around line 236 and the related collection update path. Compare the reported behavior with the Mongo and accounts-oauth call sites shown in the trace. Done means confirming the cause and defining a supported fix or documenting why the current update call must change.

Written by the indexing model from the issue text.