Add Sparsemax activation function

[zaleslaw]

The Sparsemax activation function is missed in KotlinDL.

Read more about From Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classification

The desired PR addressing this issue should include:

• Implementation of activation class named as SparsemaxActivation (you can take inspiration from the implementation of HardSigmoid as reference) added to the Activations.kt file
• Documentation of activation function
• JUnit tests in api module
• Support for export of activation function to JSON (see ModelSaver.kt)
• Support for import of activation function from JSON (see ModelLoader.kt)

P.S. If you want to take this ticket, please leave the comment below
P.P.S Read the Contributing Guidelines.
P.P.P.S. It's a difficult task, it could takes a lo time to implement and verify, maybe some ops could be missed

The reference implementation could be taken from tensorflow-addons

[cagriyildirimR]

I would like to work on this one.

[zaleslaw]

Fixed in #261