Analyze impact of warmup/tuning iterations on performance

[ahsan-ca]

HipBlasLt created a document on how to tune your own hipblas calls. https://github.com/ROCm/hipBLASLt/blob/develop/docs/how-to-use-hipblaslt-offline-tuning.rst. After checking with the hipBLASLt team, we figured our way of tuning hipBLASLt is similar to how the tool in the document describes, with the difference being the number of iterations for warmup/tuning calls.

The purpose of this issue is to analyze the impact on performance of warmup/tuning calls for a model (bert-base-cased-1) by running them a number of times in the iteration space and gathering/analyzing data.