Add a helper in math/quantum to calculate the matrix square root of sparse, hermitian matrices

[JerryChen97]

Context:
This need arose from #6963 where we realized that scipy.sparse could NOT help with a sparsity-preserved sqrtm as a sparse counterpart of its own scipy.linalg.sqrtm using Schur decomposition. Soon we realized that there could be a long list of lost functionality, e.g. eigh, sqrtm etc. We would like to update step by step such that when we implement the sparse representation for various matrices we do not have to convert them into dense repr, but this might involve quite some effort considering performance benchmark and lack of references. Therefore, we need some temporary solution that helps with specific Operator such as BlockEncode to avoid an explicit cast back to dense, thought the performance might vary.

For the BlockEncode considered in this PR, it specifically used sqrtm for a special class of matrices $Id - A^\dagger A$ where $A$ is the input matrix to encode into quantum circuits. This is very close to our own sqrt_matrix in qml.math.quantum that deals with the square root computation of a density matrix; our previous dense-matrix algorithm also made this assumption and directly called sqrt_matrix in the needed place. Hence we put forward a commonly used iterative method here suitable for similar Hermitian matrices to have square roots.

A quick introduction of what is the Denman–Beavers iteration used by this PR can be found at Wikipedia.

Description of the Change:
Provide a new helper in quantum.py

Benefits:
Enable the sparse version of BlockEncode

Possible Drawbacks:
No major. This method might have poor performance if used improperly.

Related GitHub Issues:
[sc-84693]

[github-actions]

Hello. You may have forgotten to update the changelog!
Please edit doc/releases/changelog-dev.md with:

• A one-to-two sentence description of the change. You may include a small working example for new features.
• A link back to this PR.
• Your name (or GitHub username) in the contributors section.