This is a nice measurement on B5.

Figure 1 zooms a bit more on the vertical axis to see more clearly the noise level.

On PD2 the noise floor is at 1e-7 mW/rtHz. For 16mW of power one would expect a shot noise of 7.4e-8 mW/rtHz. But there is also the shot noise of the PSTAB that is at comparable level, so in total the 1e-7 mW/rtHz is consistent with the quadratic sum of the shot noise of the PSTAB loop and the shot noise of detecting the beam on B5.

On B5P PD1, the noise floor is at 2.2e-8 mW/rtHz. This is most likely electronic noise. Relative to the power on the S-polarization it is an 1.4e-9 1/rtHz RIN. The angle of the power split between P-pol and S-pol is sqrt(0.19/16) = 0.11rad.  Using the formula from (VIR-0032A-24, slide 2). This corresponds to an upper limit on the polarization angle fluctuations of 6.4e-9 rad/rtHz.

Repeating this measurement with the polarization analyzer installed (once with the half-wave plate and once with the quarter-waveplate), should allow a better measurement of the input beam polarization angle fluctuations by a factor 6.