Update stale reversible sampling helper reference

[vtomole]

The docstring references preprocess_lcu_coefficients_for_reversible_sampling but it seems like this method was replaced with preprocess_probabilities_for_reversible_sampling.

[gemini-code-assist]

Code Review

This pull request updates a docstring reference in state_preparation_alias_sampling.py to point to the correct utility function. Feedback suggests removing a confusing and inconsistent phrase added to the docstring. Additionally, a bug was identified in the calculation of the precision parameter mu, which incorrectly uses the total number of coefficients instead of the number of non-zero coefficients, potentially leading to insufficient precision in the prepared state.

[gemini-code-assist]

The update to the function reference is correct, but the added phrase with unnormalized_probabilities is confusing and inconsistent with other docstrings in this file (e.g., line 188). In from_n_coeff, the probabilities are symbolic and not provided as a list.

Additionally, I noticed a bug in the implementation on line 466 (visible in context): mu is calculated using n_coeff (the total number of indices $L$) instead of n_nonzero_coeff (the sparsity $d$). Since alias sampling for a sparse state is performed over the $d$ non-zero coefficients, the precision \nu should be determined by n_nonzero_coeff. Using n_coeff will result in insufficient precision in the prepared state.

Suggested change

	See `qualtran.linalg.lcu_util.preprocess_probabilities_for_reversible_sampling`
	with `unnormalized_probabilities` for more information.
	See qualtran.linalg.lcu_util.preprocess_probabilities_for_reversible_sampling
	for more information.