Verbose GPU mode should report the grid size of the kernels

[e-kayrakli]

This was asked in Gitter before, but I can't find the context. Today, I wanted to confirm gpu.itersPerThread behavior in the following code:

on here.gpus[0] {
  foreach i in 1..1000 {
    // will run on 1000 GPU threads
  }

  @gpu.itersPerThread(10)
  foreach i in 1..1000 {
    // will run on 100 GPU threads
  }
}

using startVerboseGpu/stopVerboseGpu, I get

0 (gpu 0): $CHPL_HOME/itersPerThread.chpl:5: kernel launch (block size: 512x1x1)
0 (gpu 0): $CHPL_HOME/itersPerThread.chpl:10: kernel launch (block size: 512x1x1)

This is correct, but without grid size, we can't see how many threads were used for each kernel. --debugGpu is the only way to confirm that today, and --debugGpu is not supposed to be user-facing.