Nvidia

We should add an example that demonstrates how to use a CUDA stream with thrust algorithms, e.g:

thrust::reduce(thrust::device.on(stream), ...);

Describe the bug
If min_samples_split is a small float, then it may be equivalent to splitting on < 2 samples. This causes cuml to blow up:

RuntimeError: exception occured! file=../src/decisiontree/decisiontree.cu line=41: Invalid value for min_samples_split: 1. Should be >= 2.
Obtained 64 stack frames
#0 in /home/mboling/miniconda3/lib/python3.8/site-packages/cuml/common/../../..

Describe the solution you'd like

Improve this error to provide both the found hardware and the expected hardware

 No compatible device was found for instantiating TensorRT **engine**

I would like to be able to determine my device-wide primitive's temp_storage_bytes before I have all of the primitive's arguments ready. The interface for obtaining it ostensibly requires everything to be ready for the actual run - and the documentation does not make it clear what arguments it actually needs and what it's going to do with them (e.g. will it look at the input at all, in any way? Pr