CUDA Callback API (HAPI)

[mayantaylor]

• update documentation
• remove ctypes changes
• merge in corresponding charm++ pr

[ritvikrao]

Have you confirmed that the example works (e.g. on delta), since we can't really add a test for the code here?

[mayantaylor]

Have you confirmed that the example works (e.g. on delta), since we can't really add a test for the code here?

Tested the HAPI example on Delta.

[ritvikrao]

Should Numba and/or Torch be added as dependencies in charm4py for GPU builds?