find_one() returns list of objects instead of one .. when using raw query to find a specific document in an embedded array of documents

# Bug

I have a document like this:
```
{
"_id": ObjectId("user1"),
"addresses": [
{
"id": ObjectId("a1"),
"details": [....]
},
{
"id": ObjectId("a2"),
"details": [....]
}
],
}
```
I want to get the details of document `a1` only.

### Current Behavior
When using this code, it returns a document with the full embedded array
```
address = await engine.find_one(User, {"_id": ObjectId("user1"),
"addresses": {
"$elemMatch": {
"id": ObjectId("a1")
}
}})
```
It returns:
`{ "_id" : ObjectId("user1"), "addresses" : [ { "id" : ObjectId("a1"), "details" : [....]}, { "id" : ObjectId("a2"), "details" : [....]}, .... ... ] }`

### Expected behavior

When I use the mongo shell:
`db.users.find({_id: ObjectId("user1")}, {addresses: { $elemMatch: {"id": ObjectId("a1")} }})`
It returns:
`{ "_id" : ObjectId("user1"), "addresses" : [ { "id" : ObjectId("a1"), "details" : [....]} ] }`

### Environment

- ODMantic version: 0.3.2
- MongoDB version: 4.2.12
- Pydantic infos (output of `python -c "import pydantic.utils; print(pydantic.utils.version_info())"`): 1.7.3

**Additional context**

Sorry If I missed something, It's my first time working with NoSQL

Start at the engine.find_one() entry point and compare how its raw query handles the embedded addresses $elemMatch condition with the Mongo shell example. Done means a query targeting address a1 returns only that embedded address, rather than the full addresses array.

Written by the indexing model from the issue text.