Properly compare floats in gemm-block-sparse-microkernel-tester (#82947)
Similar to https://github.com/pytorch/pytorch/pull/82688
Example of current failure:
```
xplat/caffe2/aten/src/ATen/native/quantized/cpu/qnnpack/test/gemm-block-sparse-microkernel-tester.h:480
Expected equality of these values:
c[mIndex * cStride() + nIndex]
Which is: 34730.7
acc[mIndex * n() + nIndex]
Which is: 34730.7
at 7, 2: reference = 34730.65625, optimized = 34730.65234375, Mr x Nr = 8 x 4, M x N x K = 8 x 4 x 5
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/82947
Approved by: https://github.com/huydhn