Adds qBraid as a remote REST target in CUDA-Q

[TheGupta2012]

Summary

Adds qBraid as a remote REST target in CUDA-Q. Kernels authored via
cudaq.set_target("qbraid", ...) are lowered to OpenQASM 2.0, submitted to the qBraid v2
API, and results are parsed back into cudaq::sample_result — with zero dependency on
the external qbraid-sdk Python package.

What's included

• Target qbraid registered via runtime/cudaq/platform/default/rest/helpers/qbraid/

  • QbraidServerHelper.cpp — v2 API client (submit, poll, fetch, parse)
  • qbraid.yml — target config, compiler pipeline definition, declares machine and api_key arguments

• Retry-with-backoff in processResults absorbs qBraid v2's COMPLETED→result-queryable
  race.

• Specific error handling for 401 / 403 and 409 while processing job results

Testing

• C++ unit tests (unittests/backends/qbraid/): sample sync/async, observe sync/async,
  future serialization, api_key-via-target-arg, job failure, result retry, retry
  exhaustion
• Python tests (python/tests/backends/test_Qbraid.py): full Python-binding coverage
  against the mock
• Target tests (targettests/qbraid/*.cpp): 13 C++ lit tests shared with other QASM2
  backends

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[TheGupta2012]

Thank you @mitchdz for giving the review comments. Could you please approve the test runs so that the changes can be validated?

[mitchdz]

/ok to test f4907ac

Command Bot: Processing...

[github-actions]

CUDA Quantum Docs Bot: A preview of the documentation can be found here.

[TheGupta2012]

Thanks @mitchdz for approving the tests, they were helpful in figuring out a couple of bugs. I've triaged the failures and CI should hopefully pass now

[mitchdz]

/ok to test 39cfb72

Command Bot: Processing...

[mitchdz]

@TheGupta2012 I just reran CI. Sorry for the delay, I missed the email of your notification.

[TheGupta2012]

The nightly integration test failed at https://github.com/NVIDIA/cuda-quantum/actions/runs/25186596807/job/73845759395#step:6:415 with:

Warning: File not found: targettests/qbraid/*.cpp
Warning: File not found: docs/sphinx/targets/cpp/qbraid.cpp
Warning: File not found: docs/sphinx/targets/python/qbraid.py

@mitchdz thank you for running the tests. It seems like they are failing because the integration tests specific to qBraid are picking up the base cuda quantum docker image, not the one built by this PR. See the integration test log for verification

This means that it won't contain the qbraid backend and that is why nvq++ produces the invalid target error for qbraid.

Ideally, we should be passing an image built from the PR branch (migrate-cudaq-v2, head a525bfb44…) inside the cudaq_test_image input var while launching the integration tests.

[TheGupta2012]

After digging in a little with Claude I found that the CI also produced PR specific images -

A PR-specific image was pushed by Deployments run 25191512163: ghcr.io/nvidia/cuda-quantum:amd64-cu12-pr-4328-base@sha256:7519cf7d…, revision a525bfb….

Could you re-run integration tests with target=qbraid, cudaq_test_image=ghcr.io/nvidia/cuda-quantum:amd64-cu12-pr-4328-base@sha256:7519cf7d8fcb66a0f6be5123cb5b4184d76b0ee2991015ad522cf230539a36f7, and commit_sha=a525bfb447d3329dc2f47074b402aa94cfee8998?

Heads-up: the integration test setup job's regctl manifest parsing assumes a multi-arch manifest list — if the per-platform tag is a flat manifest, the setup step may need a tweak.

Not sure if this is supposed to be the target image, but hope this helps