Automatic denormalization / embedding

`populate()` makes it easy to pull in data from another collection, but in Mongoose 5.x you're on your own for embedding data.

If you're pulling a blog post's author from a 'User' collection, the below works fine.

```javascript
const blogPostSchema = Schema({
title: String,
content: String,
authorId: String,
tags: [String]
});

blogPostSchema.virtual('author', {
ref: 'User',
localField: 'authorId',
foreignField: '_id',
justOne: true
});
```

But if you want to embed `author`, it's up to you to set up middleware. Would be great if Mongoose could handle this:

```javascript
const blogPostSchema = Schema({
title: String,
content: String,
authorId: String,
author: {
type: UserSchema.pick(['name', 'email']),
embed: 'User',
localField: 'authorId',
foreignField: '_id'
},
tags: [String]
});
```

Start by reviewing the populate() and middleware behavior referenced in the issue, then trace how schema virtuals and embedded fields are defined. Compare the proposed blogPostSchema example with existing Mongoose schema and population tests. Done should mean a supported automatic denormalization or embedding workflow matching the requested configuration, with documented behavior and tests.

Written by the indexing model from the issue text.