The basic idea of ToBI comes from the work of English intonationalist in the 70's, especially Liberman and Pierrehumbert, operating within an autosegmental phonological framework. The stipulation that you have only H and L in Tobi is for the most part an arbitrary stipulation, and there is nothing wrong with positing M as a tonal element (as Liberman did). The strongest reason for not including "M" is that it is not necessary. And I should point out that ToBI is more an "approach" than a formal theory, so there is no specific principle prohibiting the addition of other categories.
However, the nature of ToBI is that it is a system for talking about categories, and not continua. Thus expanding ToBI to have numeric operators in the representation would totally change what ToBI is. You could do that, but then it would not be ToBI, it would be a new system. Similarly, ToBI is about pitch, and it isn't a general theory of all of prosody. Again, you could construct a new system that is intended to express durational properties, but then you should call it something else -- perhaps DuBI (Du for "duration).