DaskXGBRegressor with tweedie objective throws error when N-chunks = 1

I've been trying to use Dask to chunk data and train a regressor, but I kept receiving this error when trying out some sample code:

```
xgboost.core.XGBoostError: [15:26:58] /Users/runner/work/xgboost/xgboost/src/objective/regression_obj.cu:528: Check failed: info.labels.Size() != 0U (0 vs. 0) : label set cannot be empty
```

After some experimentation I realised that when I changed the `chunks` in my dask to a smaller size that the full array length, the error would go away.

I've included a snippet which should help recreate the issue:

```python
import distributed
from xgboost import dask as dxgb
import dask.array as da

if __name__ == "__main__":
client = distributed.Client()
samples = 1000
cols = 10

X_train = da.random.random(size=(samples, cols), chunks=1000) # set to

Start with the provided Python reproducer and DaskXGBRegressor.fit, then trace the empty-label failure to xgboost/src/objective/regression_obj.cu:528. Compare the one-chunk and smaller-chunk cases, including the reg:tweedie objective. Done means the supplied one-chunk example trains successfully without the label-set error.

Written by the indexing model from the issue text.