added easy-adoption examples and README

[RorryB]

Thank you for creating such a well-implemented KAN library with a straightforward interface! After discovering your repo through a recommendation at AI4, we were able to build on top of it effectively.

This PR adds integration with a module-swapping utility from my repo that we found incredibly helpful in our experiments. It enabled us to seamlessly replace nn.Linear layers with your KANLinear across multiple architectures with minimal code changes.
Benefits of this addition:

• Lowers the barrier to entry for PyTorch users wanting to experiment with KANs in existing codebases
• Accelerates experimentation by making it trivial to swap KAN layers into any architecture
• Expands your project's reach to teams who want to test KANs without major refactoring
• Expands your project's reach to teams working with pre-built models where simply replacing nn.Linear with KANLinear by hand would require greater effort.

The included README provides clear usage examples and a more detailed description. I believe this could help more practitioners adopt your work. Happy to address any feedback or make adjustments!