Clamp num_marginals to the available workload size

[vvv214]

Summary

• avoid np.random.choice(..., replace=False) failures when --num_marginals exceeds the filtered workload size
• keep the existing random subsampling behavior unchanged when enough candidates are available
• add the bound check directly in the mechanism entrypoints that materialize the filtered workload

Why

After filtering by degree and max_cells, the CLI scripts do not currently validate --num_marginals against the number of remaining candidates. There is not an earlier shared validation point before these filtered workloads are materialized, so requesting more marginals than remain raises a ValueError from choice(..., replace=False).

Testing

• ~/.local/bin/uv run --extra dev pytest -q

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.