[BUG]: Radix sort test doesn't catch allocation exceptions

[gevtushenko]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this bug and that I agree to the Code of Conduct

Type of Bug

Something else

Component

CUB

Describe the bug

Radix sort test checks available VRAM before attempting to test large problem sizes. Although the host memory requirements were recently reduced, the test still requires more host memory than the device one. We should rewrite the test in the same way the merge test is written:

try {
 // attempt to test on a large input
}
catch (std::bad_alloc &e) {
  // skip
}

How to Reproduce

Launch radix sort test with VRAM/RAM ratio < 1.

Expected behavior

Test should work on any system with supported GPU.

Reproduction link

No response

Operating System

No response

nvidia-smi output

No response

NVCC version

No response