Handling default values in QB

If a field has a default value in the schema, we no longer need to pass it to the `insert` statement. In terms of Python models this means that the field can be omitted as well. The way we want to implement this is by using default values for the Pydantic fields, so that we don't need to change whether a field is optional or required.

In EdgeQL we have `__default__` expression to refer to the defined default value. We can expose a `std.default` that will play the role of this placeholder in Python models. Thus we'd map types in the following pattern:
- `required foo: str` -> `foo: str`
- `required foo: str {default := 'bar'}` -> `foo: str | std.default`
- `foo: str` -> `foo: str | None`
- `foo: str {default := 'bar'}` -> `foo: str | std.default | None`

Notice that `None` is a separate value here that is not equivalent to "default". In general, we cannot assume that `None` must be replaced with default because this reasoning doesn't work for optional fields (where `None` is a legitimate value).

In practice, the users will have to decide how they handle the input data. For a simple case of a required field with a default and no valid "falsy" values, a simple `Spam(data.foo or std.default)` pattern might be appropriate. If there are valid "falsy" values (like `''` or 0), then a more robust pattern would be `Spam(data.foo if data.foo is not None else std.default)`. Etc.

In terms of the `save()` functionality, the fields with a value of `std.default` can simply be omitted from the `insert`.

Fields with value of `std.default`, should be omitted from other forms of serialization, because the intent is to not have a value specified and also because we cannot consistently serialize `std.default` into JSON.