Here, Tr is the time referred to (by the verb or verb group, e.g., have told, have been, told, was), and To is the time of orientation, which equates to the time of utterance in this question.
Now, my question is about the distinction made in the quoted portion of CGEL between the continuative and non-continuative readings of the perfect. If I understand it correctly, CGEL is making the distiction that the perfect locates Tr "before and up to To" and "wholly before To" in the continuative and non-continuative reading, respectively.
The following perfect in bold I think has the continuative reading:
(1) She has been writing the book since she was in her twenties and at last it's finished.
But here, Tr, the time referred to by the verb group has been writing, doesn't seem to extend up to the time of utterance (To), because at the time of uttrance, she's not longer writing the book.
Am I misunderstanding CGEL or is CGEL's distinction between the continuative and non-continuative reading of the perfect doesn't really apply to example (1)?
Here's some evidence supporting CGEL's claim that the continuative reading's Tr includes To:
A 2002 linguistics paper titled "Event Structure and the Perfect - Stanford University" by Paul Kiparsky (page 5) quotes another paper (Mittwoch 1988) to say this:
The universal reading requires an adverb specifying a duration (such as always, since 1960 or for two years)...
[T]he boundaries that define the duration are understood in an exclusive way in the existential reading but in an inclusive way in the universal reading (Mittwoch 1988). The sentence
 I have been in Hyderabad since 1977.
is false on the existential reading if I last was in Hyderabad in 1977 or if I have just landed on my first visit there; it is the intervening time that counts (exclusive boundaries). For the universal reading of  to be true I must have been there in 1977 and I must be there now (inclusive boundaries).
(Boldface mine.) (Here, the existential reading refers to the experiential reading, whereas the universal reading refers to the continuative reading.)
So what this paper is saying is that the continuative reading must include the boundaries that define the duration specified by an adverb (e.g., since 1977).
Since this 2002 paper by a reputable linguist quotes a 1988 paper to make this point, I highly doubt that this specific claim made by this paper is questionable. Moreover, this paper's claim is in line with CGEL' explanation that the continuative reading's Tr includes To.