Made api with configurable kernel parameters

[2020saurav]

Kernels for spatial as well as region was hardcoded into the library and were not configurable from the calling python functions. I have modified the library to make it configurable.

New way to call the segment function:

(seg_image, labels_image, n_regions) = pms.segment(original_image, spatial_radius=5,
          range_radius=2, min_density=50, skernel="Gaussian", rkernel="Uniform")

Note: Specifying the kernels is optional. Default for both is set to be Uniform kernel.