You are correct. The problem here is not the concept of estimating, but the fact that they are teaching kids to estimate improperly. The proper technique is to pick the degree of accuracy and then round to the nearest zero. In this case, the most accurate estimate is achieved by rounding up or down to the nearest zero: 354+291=645 becomes 350+290=640.
I would be interested in seeing more of this work book. Are they teaching truncation as the only method, or is this the first day and then the next day they will show that rounding is more accurate than truncating?
Some of my complaints with some of the newer math techniques are that they get stuck on more explanatory but less efficient methods (like lattice multiplication, if I have the term right) rather than just using that as a stepping stone to the best methods.