* llvm-flang Only build offload code if cuda enabled
The current version executes `cmake(*args)` always as part of the post install. If device offload is not part of the build, this results in referencing `args` without it being set and the error:
```
==> Error: UnboundLocalError: local variable 'args' referenced before assignment
```
Looking at prevoous version of `llvm-package.py` this whole routine appears to be only required for offload, some indent `cmake/make/install` to be under the `if`.
* Update package.py
Add comment
* Modify Flang NVidia GPU variant to make use of built-in CudaPackage
* Add OpenMP Offload patch if March 2019 compiler is selected.
* Flang parallel build has a race condition.
* llvm-flang now uses built-in CudaPackage.
* Add variant for different build releases.
* Fix OpenMP target offload for NVidia GPUs.
* Additional commong flags that are needed with comments.
* NVidia BC required for libomp target requires special treatment. Use clang built in previous step to re-compile libomptarget.
* Add 20181226 release of pgmath
* 20190329 release of pgmath requires match for CMPLX macro.
* Add llvm-flang package for Flang fork of LLVM
* Add new and old flang releases.
* Add cmake and python dependencies.
* Update dependencies on llvm-flang and pgmath.
* Fix cmake args and change spec to reflect llvm-flang package.
* change copyright date through 2020.
* Reference Flang package more explicitly.
* More robust support for python executable.
* import os no longer needed, picked up by flake8.
* Use built-in target spec. Variant and targets follow style in main LLVM package.
* Get rid of targets list and only support one target for now.
* Sparc does not appear to be supported in Flang.
* Raise InstallError if architecture not supported.
