Missing factor 2 for test-mass noise

As arguments to the test-mass acceleration noise generator function, we give the spectral parameters for the jitter motion of the test-mass. Internally, we then convert this test-mass motion to equivalent fractional frequency deviations of the beam reflected off the test mass.

We multiplying by 1 / (2π f c)^2, but we are missing a factor 2 to account for the two-way optical path. We account for this in LISANode, and this yields an obvious deviation from LISANode and the MRDv1 at low frequencies [credit to @maudelejeune]:

We should add this missing factor 2, and document it.