Module layout and use of Step classes on new contributions

[dannymeijer]

Recently, we've been getting quite a few PRs opened to extend Koheesio with various Transformation style functions.
Unfortunately, all of these PRs (#118, #119, #122, #123, #127, and #128 so far) are not following the intended code style of Koheesio.

Since these PRs are very similar in nature, I want to use this discussion to come up with a proper strategy on how to incorporate these changes properly.

What I would like to see for these contributions is:

• Use Step classes instead of plain functions
  • We could probably argue that most of these changes are actually similar to the Transformation class we have in the spark implementation; let's discuss on how to incorporate that - I can help adjusting/generalizing the Transformation class similar to what we did for Reader.
• Documentation style used should be NumpyDocs - not Google style
• Make sure every new module has proper docs including examples. Add module docstrings to explain what your module is doing.
• We need to find a proper spot to place the new modules; the koheesio root is not meant to be used as these PRs are trying to. Instead, if this is pandas for example, we should place this in the pandas module and adjust the code accordingly.

Please let me know if and how I can help. Would love to do a zoom call with you to explain or even pair program.

[dannymeijer]

add #131 to the list of the PRs I mentioned prior.

[dannymeijer]

I opened #141 to help make life easier. Once this gets merged, you can use the pandas based Transformation class to restructure your code.