[rocm-libraries] ROCm/rocm-libraries#4791 (commit 6cc17c6)

[CK][CK TILE] Improve oob check

## Motivation

Improve OOB checks. Remove permutes which have been generated by thread
buffer zero clear. at now in assembly there is only condmask instead of
permute + condmask.

Change number of KPack for generated instances

## Technical Details

Remove permute instructions from assembly

## Test Plan

test_grouped_convnd_fwd_tile

## Test Result

passed

## Submission Checklist

- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

profiler/src/profile_grouped_conv_fwd_tile.cpp

@@ -96,7 +96,7 @@ int call_profiler(const ckt::Args<SIGNATURE>& args, bool time_kernel)
     std::string op_name;
     bool valid;
     std::tie(valid, avg_time, op_name) = ckp::run_grouped_conv_forward_tile_algs(
-        args, inputs.get(), outputs.get(), ck_tile::stream_config{nullptr, time_kernel});
+        args, inputs.get(), outputs.get(), ck_tile::stream_config{nullptr, time_kernel, 0, 5, 50});
     if(time_kernel)
     {
         std::cout << "Best configuration parameters:" << "\nname: " << op_name