Using the the Least-Square computed covariance matrix as input in Kalman: about measurement parameters

Hi everyone,
During an OD process, I use a Least Square estimator to compute a covariance matrix, which I then use as input covariance matrix in a Kalman estimator.
When only 6 orbital parameters and some propagation parameters are estimated, the computed covariance can be transfered directly to the KalmanEstimatorBuilder with the addPropagationConfigurationMethod.

But when some measurement parameters are also estimated, I should use the estimatedMeasurementsParameters method, in which the covariance deals only with the measurement parameters. In that case, I suppose I can extract the measurement part of the LS computed covariance, but then I seem to loose some information through the loss of the non diagonal elements dealing with the measurement parameters.

What is the correct way of doing it ?

I should mention that in some tutorials, I found the covariance matrix given with the estimatedMeasurementsParameters method is null, which confuses me further…

Additionally: why does the BatchLSEstimator getPropagatorParametersDrivers and getMeasurementsParametersDrivers methods both offer the possibility to get all drivers or only estimated ones, whereas the AbstractKalmanEstimator same methods only allows this choice for propagation drivers and not for measurements drivers ?

Thanks in advance !