thanks for your interest in the MMB!
In general, the Rudebusch and Swanson (2012) model is a perfect match for our database. The most work-intense part would be to correctly replicate it in Dynare format (the documentation in the end of the paper provides help for the equations part).
Regarding your question on the model specific rule: The implementation of the model specific policy rule is supported in the MMB, please check the documentation for further details. For the comparison, however, you just need to replace it with the common policy rule (details on how to do this can be found in our user guide).
Regarding your question on the model ouputs. For sure, it is then possible to check for any variable (also the term premium). You can find the respective results.mat files in the model folder after simulation. As most models in the MMB are linearized, changing the order of approximation would simply have no effect. For your own comparison, however you can change the order from 2 to 3, to do that please consult the stoch_simul_MMB.m file in the MMB_OPTIONS folder.
Unfortunately, creating an adaptive learning version of the model is not trivial and cannot easily be explained here. We invite you to study the AL model files and the respective literature for further reference. However, the first step is always to replicate the model. As soon as you have a running Dynare replication, and you experience trouble implementing it, please feel free to send us your codes and we can provide you further support.
We hope this helps!