Improve FRF estimation by using GraFIT method.

Improve FRF estimation by using the _Local Polynomial Method_, as also detailed in the Pintelon and Schoukens textbook. An example, specifically tailored for gravitational-wave detectors, is documented in the [2025 GraFIT paper](https://arxiv.org/abs/2503.17084v1). There is an accompanying [Matlab code base](https://github.com/MathynVanD/GraFIT) and which has also been translated (with minimal bug fixing) to Python.

Swapping to this FRF method is expected to lead to equal SNR measurements with less data collected - consequentially saving time.

Note that the formalism, equations 16 and 17, in the arXiv version are slightly erroneous. Specifically the loop suppression term (I + K * G)^-1 is in the wrong position - in the paper this is written as pre-multiplication on the RHS but it should be post-multiplication instead. I have, last year, communicated this error to the paper authors.

No repository files or tests are named. Start by reading the current Python FRF estimation implementation, then compare its formalism with the 2025 GraFIT paper and accompanying Matlab code, including the correction to equations 16 and 17. Done means the estimator uses the Local Polynomial Method and demonstrates the expected SNR or data-efficiency improvement.

Written by the indexing model from the issue text.