In order to improve the numerical accuracy of the signals that go through the TypeA suspensions (in particular the ETMs), I have added an option to make the offloading heirarchical.  I have also kept the option to have the suspensions operate in the way that they have been operating, to ensure a smooth transition to the new architecture. 

The new heirarchical scheme means that, rather than taking the output of ISCINF and sending that directly to the inputs of the LOCK filter banks for all 3 payload stages (TM, IM, MN), we now can take the output of a stage, and send it to the input of the next stage.  For example, now the output of the TM_LOCK filters can be passed to the input of the IM_LOCK filters.  This means that the filter shaping only needs to be that required for making a stable crossover between the stages, and it should only need to go up like f.  In the old way, the IM stage filters would need to go up like f^2, and the MN filters would need to go up like f^3.  Another benefit of the heirarchical architecture is that we can change the shape of the DARM loop in only one filter bank, and the result will be propagated to all stages, and the crossovers between the stages will continue to be stable.  In the old parallel architecture, you might have to check the filters and crossovers of each stage to ensure stability, so this new scheme should save some effort later. 

In order to pass some signals up the chain without actuating on a given stage (eg, we want to actuate on TM and MN stages, skipping the IM stage), I have also added extra on/off switches after the lock filter banks. 

Nakano-kun is working on making the screens for the TypeA suspensions reflect the new options. To use the old architecture, set K1:VIS-[optic]_HEIR_SWITCH = 0, to not use the heirarchical scheme.  If you want to use the heirarchical scheme, set the switch to 1.