The real question that should be asked is, why would anyone write a textbook in the first place? I will only address the question as it applies to phonology, but similar answers can be given for all domains of linguistics. First, there is a difference between "making reference to" and "making extensive use of". I could mention that Lushootseed has only 1 or two words with nasal consonants, but I can't make extensive use of Lushootseed in illustrating some important point about phonology, nor can I do that with Makah or Kutenai.
Every textbook author has to make a decision as to what "the point" is. Some authors favor unloading all information they can find in a domain. One (generative-oriented) perspective on phonology is that phonology is about computing changes from underlying to surface form. Not all languages have interesting changes from underlying to surface form, so Lushootseed for the most part simply assembles morphemes and pronounces the result. One phonological change is the change /txʷ,dxʷ/ → [tu,du] in some context (involving verbs – as an exercise, look at the available materials to figure out where this happens), but that is a morpheme-specific alternation, so not necessarily of any use in a textbook (depending on the degree of generality that, as an author, you demand of your processes). Unless the point is to exemplify "consonants becoming vowels", this fact doesn't exemplify anything. Then finally, there is the problem of actually showing in a textbook-suitable fashion that this even happens. Paradigms are really important for some perspectives (esp. as exemplified in Kenstowicz's book).
First, as an author, you have to have a clear understanding of the standards you want to convey. Some authors are willing to pick up a word or two from a source and simply assert "/kʷaxʷadxʷ-bš/ → [kʷaxʷadubš]", but other require more data and more convincing. This brings us to the problem of descriptive grammars. There is lamentably little by way of grammatical description of Lushootseed: a couple of dissertations, a textbook (possibly out of print), a collection of data hosted by John Lawler, a few journal articles, and a few people who worked with the last fluent speakers (and whatever unpublished notes they have). Kenstowicz does not know Lushootseed at all, so it would be ridiculous to expect him to know of this process and be able to work it into a textbook. I know a bit of Lushootseed, and know that my hard-core data on this process is limited; and I also know that the process doesn't illustrate something important that needs to go in a textbook.
When it comes to Makah, I know virtually nothing about the language, though I think there could be some interesting phonological alternations (figuring that Makah is similar to Nootka) – but I don't know of a Makah grammar that I could peruse to pull out paradigmatic examples. As for Kutenai, I know nothing.
There are similar cogent reasons for the lacunae that you call out in Kenstowicz's textbook. You might want to engage in a larger-scale comparison of textbooks, at a minimum comparing K's textbook with earlier Kenstowicz and Kisseberth, to see what languages are employed, for what purpose. Tupi, being an extinct language with inadequate synchronic description, is not going to figure in phonology textbooks that have high demands of paradigmicity (where the facts demonstrate what the underlying forms must be, and URs aren't just handed down from on high). There is limited potential with related Guarani, where there probably is something about nasal spread that is interesting: but the data simply have not been there for the average work-a-day phonologist. The one thing that might be of some interest is the question of whether there actually is a phonological rule of nasal spreading. There is reasonable evidence that the claimed nasal spread is low-level phonetic coarticulation, which is a question that has to be addressed experimentally with a large supply of speakers, not written-language assertions. Similar issues thwart attempts to incorporate Aztec (Nahuatl), Quechua (a language family) and Maya (a language family) into a Kenstowicz-like textbook.
I also suggest paying more attention to authors' data sources. Kenstowicz has worked on a number of Asian and African languages, and has had students who did so, therefore those languages figure prominently in his textbook. If he had worked extensively on Lakota or Ojibwa, he might have included paradigms from those languages. It turns out that the phonologists who have worked on those languages mostly haven't written textbooks, or have written textbooks of a completely different nature. The fascinating phonology of many Athabascan languages is simply too complex to be of use in a textbook, unless the book intends to use Navaho or Slave as the main language to exemplify all phonological concepts (any takers? E.g., convert Rice's Grammar of Slave into an introductory phonology textbook).
In short, if an author has no personal knowledge of a language, little reasonable chance of gaining adequate knowledge, and no concrete reason to believe that there is a pedagogical payoff in the language, that language is unlikely to figure in a textbook.
Once you move past textbooks and look at phonology at large, your claim is simply wrong. Native American languages have figured prominently in theoretical phonology – see Ojibwa, Menominee, Lakota, Sarcee, Navaho, Klamath (the mother-lode language), Karok, Tunica, Choctaw, Paiute, Mohawk. Salishan languages mostly don't figure in discussions of phonology, except for reduplication studies, but they do figure prominently in morphosyntax and semantics.
