Fix some issues with the GPU support to enable simple Arkouda builds

[e-kayrakli]

This branch makes bunch of small-ish changes in an attempt to get Arkouda to
compile with the GPU locale model. I can separate this into multiple PRs, but I
want to test it some more as a whole. Fixes with this PR:

• Resolves Clang error compiling included C++ headers with GPU locale model #19754
  • Stops adding extern "C" in the C headers included at the command line.
    This was proposed by @milthorhpe.
  • Fixes an interop test broken by this.
• To be able to do that makes runtime's generated code interface for complex
  numbers C-based even if we're compiling with C++. (A prior PR that made some
  improvements in that direction is
  Adjust runtime headers to better handle C++ compilation #16847)
• This enables defining c_string_to_complex* functions to be included in the
  generated executable, because their return types are no longer
  std::complex<> which you can't link with C linkage.
• Adds a no gpu codegen pragma to thwart gpuization within some internal
  functions. Currently this only applies to:
  • chpl__initCopy_shapeHelp: Has PRIM_ASSIGN that gets normalized after
    gpuization, but not resolved. We probably need this function, and the lack
    might be preventing us from using loop expressions to initialize arrays on
    gpus.
• Fixes .name implementation of the GPULocale type.
  • Adds a relevant test.
• Adds support for BitOps module.
  • Adds a relevant test.
• While there, cleans up gpuTransforms.cpp a bit.

Test:

• gpu/native
• gpu/interop
• make check + spot checks in types/complex with intel as the host+target compiler
• make check + spot checks in types/complex with cray as the host+target compiler

[e-kayrakli]

@stonea -- I did the refactoring we discussed online and some more. This commit has them (with a new function this PR adds). Let me know if you think I went too far and want me to split that commit into its own PR.

@mppf -- This commit has the core of the complex-related change in the runtime. I'll do some portability testing there. Do you have any specific concerns as to the configs where what I am doing can be problematic? In your #16847, you just did a full local testing, it looks like.

[mppf]

@e-kayrakli - I don't have any concerns or specific test configurations to bring up.

[e-kayrakli]

@milthorpe -- This still needs some work, but we'll merge it soon. I'm wondering if you can give it a try to see if you can get a local, no-module Arkouda server built with it?

[stonea]

@stonea -- I did the refactoring we discussed online and some more. This commit has them (with a new function this PR adds). Let me know if you think I went too far and want me to split that commit into its own PR.

I don't think we need to be sticklers for separating refactoring changes out into a separate PR. Having it as a separate commit is nice for software archeology and reviewing purposes so I appreciate that.

[milthorpe]

Today I re-tried compiling Arkouda and got this error, which I believe is a known problem:

In file included from <built-in>:3:
In file included from /tmp/chpl-milthorpe.deleteme-jUa4RL/command-line-includes.h:7:
In file included from /noback/milthorpe/chapel-1.27.0/modules/packages/ZMQHelper/zmq_helper.h:26:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/qio/qio.h:26:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/qio/qbuffer.h:39:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/qio/deque.h:44:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/chpl-mem.h:30:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/arg.h:24:
In file included from /noback/milthorpe/chapel-1.27.0/runtime/include/chpltypes.h:39:
In file included from /noback/milthorpe/chapel-1.27.0/third-party/llvm/install/linux64-x86_64/lib/clang/14.0.0/include/cuda_wrappers/complex:33:
/auto/software/swtree/ubuntu20.04/x86_64/gcc/11.3.0/lib/gcc/x86_64-pc-linux-gnu/11.3.0/../../../../include/c++/11.3.0/type_traits:44:3: error: templates must have C++ linkage
  template<typename... _Elements>
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/chpl-milthorpe.deleteme-jUa4RL/command-line-includes.h:2:1: note: extern "C" language linkage specification begins here
extern "C" {
^

[e-kayrakli]

@milthorpe did you get that with this PR applied on main? It is not really a must that you do it for this PR to proceed, but if you are getting this error with this PR applied, that's surprising.

[milthorpe]

@milthorpe did you get that with this PR applied on main? It is not really a must that you do it for this PR to proceed, but if you are getting this error with this PR applied, that's surprising.

Oops, I didn't correctly apply the PR. With the PR, I get:

$CHPL_HOME/modules/internal/DefaultRectangular.chpl:1494: error: GPU support does not currently allow nested kernel launches. Do you have
nested forall/foreach loops or looping over a multidimensional domain?

[e-kayrakli]

Ah, OK. That's good enough for this PR, I think/hope.

FWIW, the error you are getting is also a bit surprising but probably not from the GPU support perspective. Do you have a module in your ServerModules.cfg that supports multidimensional arrays in Arkouda? (which I know is something @bmcdonald3 has recently worked on). In any case, this PR causes issues with bunch of other Arkouda modules, too. But you should be able to compile a local, no-module Arkouda server.

[bmcdonald3]

Had a chance to look into this a bit and I confirmed that the error you are running into is from the HDF5MultiDim module:

$CHPL_HOME/modules/internal/DefaultRectangular.chpl:1494: error: GPU support does not currently allow nested kernel launches. Do you have
nested forall/foreach loops or looping over a multidimensional domain?

So that can be removed by commenting out/deleting that module from the ServerModules.cfg file in the Arkouda repo.

I was able to get Arkouda to compile with GPU support using this patch with all the modules commented out from the build, but it does seem that some of the Arkouda modules still don't build with GPU support unfortunately. There is some documentation about the modular building of Arkouda located at https://github.com/Bears-R-Us/arkouda/blob/master/MODULAR.md, but I am happy to offer more assistance there if it would help.

[e-kayrakli]

Thanks for checking @bmcdonald3!

but it does seem that some of the Arkouda modules still don't build with GPU support unfortunately

That's right. I looked at only a few of those modules. It was relatively easy to find the "offending" foralls. And I believe trying to get all the modules to compile is a good stress test for our GPU support down the road.