[mpact][benchmark] Add new benchmarks

[yinying-lisa-li]

1. add feature_scale and normalization benchmarks
2. import existing kernels.

[--------------------------------- feature_scale ---------------------------------]
                                        |  torch-sparse-eager  |  torch-dense-eager
1 threads: ------------------------------------------------------------------------
      dense_([32, 32],)_float64_0       |         77.1         |         76.8      
      dense_([32, 32],)_float64_0.5     |         75.4         |         74.9      
      dense_([32, 32],)_float64_0.9     |         75.0         |         75.7      
      dense_([32, 32],)_float64_0.99    |         76.1         |         74.7      
      dense_([64, 64],)_float64_0       |         85.4         |         86.5      
      dense_([64, 64],)_float64_0.5     |         84.8         |         86.8      
      dense_([64, 64],)_float64_0.9     |         84.8         |         83.7      
      dense_([64, 64],)_float64_0.99    |         82.3         |         82.8      
      dense_([128, 128],)_float64_0     |        119.2         |        118.3      
      dense_([128, 128],)_float64_0.5   |        124.5         |        119.3      
      dense_([128, 128],)_float64_0.9   |        120.2         |        118.3      
      dense_([128, 128],)_float64_0.99  |        115.9         |        116.8      

Times are in microseconds (us).

[--------------------------------- normalization ---------------------------------]
                                        |  torch-sparse-eager  |  torch-dense-eager
1 threads: ------------------------------------------------------------------------
      dense_([32, 32],)_float64_0       |        103.0         |        101.8      
      dense_([32, 32],)_float64_0.5     |        104.4         |        100.0      
      dense_([32, 32],)_float64_0.9     |         99.4         |         99.9      
      dense_([32, 32],)_float64_0.99    |         98.5         |         97.6      
      dense_([64, 64],)_float64_0       |        113.3         |        114.1      
      dense_([64, 64],)_float64_0.5     |        113.9         |        114.9      
      dense_([64, 64],)_float64_0.9     |        117.6         |        114.2      
      dense_([64, 64],)_float64_0.99    |        116.0         |        114.2      
      dense_([128, 128],)_float64_0     |        178.8         |        178.1      
      dense_([128, 128],)_float64_0.5   |        189.4         |        174.9      
      dense_([128, 128],)_float64_0.9   |        177.7         |        189.4      
      dense_([128, 128],)_float64_0.99  |        175.3         |        167.0      

Times are in microseconds (us).