As far as searching, the index at the back of a book is likely more useful than full-text search. Usually I'm not interested in a particular term, but 'that thing that we covered last week, with the Fourier Transform', or something like that. There aren't any keywords, regexes won't help - you need a well-organized textbook where the headings and sub-headings facilitate this kind of searching.
Curriculum changes seem like the wrong way to go as well - if a tablet isn't good for teaching a subject, don't jam the subject into a format the tablet can understand. This is clearest in the testing case, but it's generally applicable: the tablet, as a device, exposes an interface. Paper exposes the ultimate interface: it accepts any free-form input. At best tablets can work in a similar way, but we have a hard time doing anything useful with this input. As we normalize the input to make it easier to process, we move away from a general purpose input device, and we force students to think and express themselves in the specific way we designed the interface to work.
While improvements to a tablet approach can resolve all of the problems you're mentioning, a regular book will never be regex searchable. A regular piece of paper will never let the teacher know your scores immediately so she can help you past problems and provide instant feedback. A regular set of class books will never decrease the need to linearly increase the number of pieces of paper needed for the amount of materials you want the students to have.