- hil::symmetric_encryption
  - provide distinct traits for Ctr/CBC modes so we can be sure these
    hardware features exist at compile time
  - allow distinct src/dst encryption or in-place operation
- fix up sam4l::aes to match the new interface, and track input/output
  with the hardware properly
- add TakeCell.map_or_else()
- tests are in boards/imix/src/aes_test.rs, and may be run by adding
  this just before the call to `kernel::main()`:

    aes_test::run();

NOTE: this commit removes capsules::symmetric_encryption, the userspace
interface to AES.  Future work will have to patch that back onto the
modified HIL, and perhaps also provide a mux for userspace+kernel users.

…'s not pretty.

These methods now have the same semantics as put/take for a TakeCell,
except that when the hardware is busy (between a call to crypt() and the
following invocation of crypt_done()), they have degenerate behavior:
put has no effect, and take returns None.

This changes the interface once more so that there is
no opportunity for confusion over the state of buffer
ownership.

- hil::symmetric_encryption
  - provide distinct traits for Ctr/CBC modes so we can be sure these
    hardware features exist at compile time
  - allow distinct src/dst encryption or in-place operation
- fix up sam4l::aes to match the new interface, and track input/output
  with the hardware properly
- add TakeCell.map_or_else()
- tests are in boards/imix/src/aes_test.rs, and may be run by adding
  this just before the call to `kernel::main()`:

    aes_test::run();

NOTE: this commit removes capsules::symmetric_encryption, the userspace
interface to AES.  Future work will have to patch that back onto the
modified HIL, and perhaps also provide a mux for userspace+kernel users.

…'s not pretty.

These methods now have the same semantics as put/take for a TakeCell,
except that when the hardware is busy (between a call to crypt() and the
following invocation of crypt_done()), they have degenerate behavior:
put has no effect, and take returns None.

This changes the interface once more so that there is
no opportunity for confusion over the state of buffer
ownership.

* The CBC-test fails because it is not supported. I suggest to change
the test structure and the trait bounds accordingly

* Because the AES tests depended on trait bounds on both AES128Cbc and
AES128Ctr => both the traits needed to implemented
* Since NRF5x only support CTR ended up in test failed
* That's why I splitted up them (I have not verified that tests are
correct though)
* The test passed on NRF5x

Add nik's changes to ccm branch so that we can make sure everything works

…ranch