Hi,
Here is some information regarding your questions:
1. You can very easily apply marks in all cross-sections: mark 1 represents the left end of the active area, and mark 3 the right end of this area. So, the parts of the cross-section which are located to the left of mark 1 or the right of mark 3 are not used in the simulation (just like if there were very high vertical walls above these marks. This is very convenient if you need to exclude a part of a cross-section when you know it won't be flooded.
In case this side part of the cross-section is to be flooded at some point in time during the simulation (e.g. a storage area during a flood event), you therefore need to describe this additional area with another calculation point. You can for example use a link channel (describing an overbank spilling) with the storage area located downstream this link, or you can use a second branch to model the flow through the side part of the cross-section (this second solution is relevant if you need to model parallel flows, e.g. in the main riverbed and in the floodplain). In both options, you get two different calculation points for the same cross-section, which will give you different levels in the main part of the cross-section and in the side part.
2. I suppose the number of iterations you're looking for is the parameter 'NoIter', available in the 'Default values' tab in the *.hd11 file. The default value of 1, which means that MIKE 11 performs 1 additional iteration after the default calculation (or 1st iteration), is usually suitable.
3. There is no easy way to remove all interpolated cross-sections at the same time. If you have much more interpolated cross-sections than 'real' surveyed sections, you can consider doing the following: select the cross-sections you want to keep, export them to a text file (File > Export), delete the whole river branch, import the saved cross-sections (File > Import) again.
If you simply added the interpolated cross-sections to get a finer resolution (with additional computational points), you can simply use the 'Maximum dx' parameter instead. This parameter is defined in the branch properties, and defines the maximum distance between calculation points for each branch. The necessary additional computational points are added at run time only, and hence it is extremely easy to revert back to a coarser resolution if needed.
I hope this is clear, otherwise just let us know.
Best regards
Mathieu