Use o3d.geometry.KDTreeFlann with o3d.t.geometry.PointCloud

### Checklist

- [X] I have searched for [similar issues](https://github.com/isl-org/Open3D/issues).
- [X] For Python issues, I have tested with the [latest development wheel](http://www.open3d.org/docs/latest/getting_started.html#development-version-pip).
- [X] I have checked the [release documentation](http://www.open3d.org/docs/release/) and the [latest documentation](http://www.open3d.org/docs/latest/) (for `master` branch).

### My Question

Hello,

I register two point clouds and then I'd like to find several nearest neighbors in the moving point cloud for each point in the fixed point cloud. I handle point clouds using `o3d.t.geometry.PointCloud`, especially because I have additional custom attributes of the points that I need to read and write.
I try to use `o3d.geometry.KDTreeFlann ` like this:

```
pcd1 = o3d.t.io.read_point_cloud("point_cloud_1.pcd")
pcd2 = o3d.t.io.read_point_cloud("point_cloud_2.pcd")
pcd1_tree = o3d.geometry.KDTreeFlann(pcd1.point.positions.numpy())
example_point = pcd2.point.positions.numpy()[0]
n, idx, d = pcd1_tree.search_hybrid_vector_3d(example_point, 10, 1)

```
which gives
`
RuntimeError: search_hybrid_vector_3d() error!
`
I assume that the data type of the point is wrong (it's `float64` numpy array) but not sure how to fix it.
Is there a way to do the same without going back to the legacy geometry classes?