v0.2.2

Release Notes

Changed

• Improved the code performance by creating a global thread pool that is
  shared across all functions that run in parallel. This avoids creating
  a new thread pool for each function call, which was causing a performance
  overhead.
• Improved the performance of the gaussian_smoothing function by using
  a more efficient algorithm to compute the convolution of the input
  data with a Gaussian kernel.