Eagerpy

New Features

* `Tensor.tensor` is now called `Tensor.raw`
* added type hints to basically all methods and framework functions
* refactored large parts of the code
* many other bug fixes and improvements following from that

New Features

* `TensorFlowTensor` now supports automatic casting to the common `dtype` for all comparison operators
* other improvements and bug fixes

New Features

* added `ep.utils.get_dummy`
* `TensorFlowTensor.sum` now automatically casts boolean to int (like `PyTorchTensor` and others)
* other improvements and new tests

New Features

* added `__mod__`
* fixed `ep.mean` (was `sum` instead of `mean`)
* reorganized the tests
* completed our test suite to reach **100% test coverage**

New Features

* **99% test coverage**
* all tests rewritten from scratch
* removed all framework-specific duplication
* added nice helper functions
* truely lazy imports of the different frameworks
* added *magic* framework wrappers that automatically apply `ep.astensor` and preform lazy imports
* `ep.torch`
* `ep.tensorflow`
* `ep.numpy`
* `ep.jax`
* added `pow()`
* added `log_softmax()`
* added `kl_div_with_logits()`
* **lots of bugfixes and improvements**

New Features

* added support for **automatic differentation** of arbitrary functions that take and return EagerPy tensors
* `value_and_grad` (function and method)
* `value_aux_and_grad` (function and method)
* `value_and_grad_fn` (function)
* added `crossentropy`
* improved the `__repr__()` shown for EagerPy tensors
* other improvements and bug fixes