I think it's wise to implement your own apps like the samples are implemented. After all, it's the Spring team that publish them, and they ought to know "best practice"
I don't fully understand your question. The command object is the model-part of "Spring MVC web layer" way of doing it. Why would you store an id or request parameter in the command object??? Seems like an oxymoron. Isn't it better to populate the comkmand object in formBackingObject()-method using the requestparamethers at hand?
So... do it the way the sample apps indicate.
"Murphy was an optimist"
(The O'Toole commentary on Murphy's Law)