Created by: FrostML
Add inference transpose_flatten_concat_fuse_pass tests for GPU and TRT to check the number of output and precision on GPU and enable TRT. CPU do not run this pass.
