Jit link

See merge request !2

Moved invocation of _locate_cuda_libdir() from constructor to add_stdlib(), it should be only called when actually needed
Renamed JitLinkModule to DynamicModule

Changed TestDriver.test_recursive_launch() in test/test_deriver.py to use pycuda.compiler.DynamicSourceModule, removed xfail marker
Added a test case for pycuda.compiler.JitLinkModule: compiles and links two .cu files.
Removed test/test_jit_link.py, all further tests in test/test_deriver.py from now on
Removed bad "import pycuda.autoinit" (how did this even get there?)

Using Context.set_limit() now.
Needed a missing limit enum CU_LIMIT_DEV_RUNTIME_SYNC_DEPTH (in CUDA8/v8.0/include/cuda.h[975], available since 3.5), added CU_LIMIT_DEV_RUNTIME_SYNC_DEPTH and CU_LIMIT_DEV_RUNTIME_PENDING_LAUNCH_COUNT (same) to pycuda.driver.limit.

+ CudaModule
  + SourceModule
  + JitLinkModule
    + DynamicSourceModule

Splitted the "old" class SourceModule in two: class CudaModule and the "new" SourceModule.

- CudaModule is now the common base class for module loading and provides common methods. All methods here were moved from the old SourceModule.
- The "new" SourceModule's interface and system requirements are 100% unchanged, it will work under all previous configuration scenarios.
- JitLinkModule requires at least CUDA 5.5 and Compute Capabilty 3.5 (that is now guarded in the constructor), it's the swiss-army-knife for non-trivial linker invocations.
- DynamicSourceModule is a special case of JitLinkModule, it exposes the same interface as SourceModule but enables dynamic parallelism (it comes with one extra optional argument in the constructor, cudalib_dir). It's meant for the trivial cases where the user has a single source file, as before with SourceModule.

So if a PyCuda user only wants to activate dynamic parallelism, all that's required is to replace "SourceModule" with "DynamicSourceModule" given that we're able to locate "cudadevrt" automagically in method _locate_cuda_libdir(), otherwise the caller must provide the CUDA library path manually in constructor argument "cudalib_dir". I do not think this can be reduced any further.

Other changes in class JitLinkModule:
- Made all add_* methods and the link() method return self
- Moved CUDA library path detection logic into method JitLinkModule._locate_cuda_libdir(), gets called only once from constructor

Added method JitLinkModule.add_stdlib() for static linking of CUDA standard libraries like "cudadevrt".

Added adaption of NVIDIA's "cdpSimplePrint - Simple Print (CUDA Dynamic Parallelism)" sample that demonstrates dynamic parallelism with a recursive kernel

Added Python class pycuda.compiler.JitLinkModule, C++ class Linker and Boost.Python wrappers for class Linker and enum CUjitInputType.

let setup.py detect CUDA_ROOT and lib directory for macOS

GPUArrray --> GPUArray

Fixed a confusing typo in the docs, where the class name "GPUArrray" was spelled incorrectly.