[Bug] Rust-backed QDP file loader ignores the configured batch limit

[viiccwen]

What

QuantumDataLoader.batches(total, size) documents total as the maximum number of batches emitted for file sources. The Rust-backed file loader ignores that value and consumes the entire file because the Python wrapper always passes batch_limit=None to the native loader.

Steps to reproduce:

import numpy as np
from qumat_qdp import QuantumDataLoader

np.save("vectors.npy", np.eye(4, dtype=np.float64))

loader = (
    QuantumDataLoader()
    .backend("rust")
    .qubits(2)
    .batches(1, size=2)
    .source_file("vectors.npy")
)

print(len(list(loader)))

Actual result:

2

Expected result:

1

Iteration should never emit more than the configured number of batches for file-backed sources.

Environment

• OS: Linux
• Python version: 3.12.12
• Qumat version: 0.7.0.dev0 (qumat-qdp 0.3.0.dev0)
• GPU: NVIDIA GeForce RTX 4090