Explicitly construct Span<EValue*> from EValue*[] in tests

[rascani]

Summary

GCC exhibits inconsistent behavior where it applies array-to-pointer decay when calling through function pointers, but not for direct function calls, even though both should invoke the same conversion to the parameter type.

Minimal reproduction:

template <int N>
struct S { S(int (&)[N]) {} };

void f(S<3>) {}

int main() {
    int a[3];
    f(a);       // OK on both GCC and Clang
    (&f)(a);    // OK on Clang, ERROR on GCC
}

A similar issue is tracked at: GCC Bugzilla 114812

In ExecuTorch, OpFunction is defined as:

using OpFunction = void (*)(KernelRuntimeContext&, Span<EValue*>);

When tests call (*func)(context, stack) where stack is an EValue*[] array, GCC decays it to EValue** before considering Span's array constructor template<size_t N> Span(T (&)[N]), causing compilation to fail.

This PR updates call sites to work around this issue by explicitly constructing the Span. An alternative option would be to convert the C arrays to std::arrays, which would not be subject to array to pointer decay.

Test plan

# Using GCC 11.5
./test/run_oss_cpp_tests.sh

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16520

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 Cancelled Job, 1 Unrelated Failure

As of commit 1cee1dc with merge base 7fa1aaa ():

CANCELLED JOB - The following job was cancelled. Please retry:

• trunk / test-models-macos-cpu (llama2, xnnpack-quantization-delegation) / macos-job (gh)
  ##[error]The operation was canceled.

UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:

• pull / android / run-emulator (gh) (#16137)
  Timeout waiting for emulator to boot.

This comment was automatically generated by Dr. CI and updates every 15 minutes.