"Juncture" is an analytic entity constructed to account for problems in phonemicization, which primarily encodes morphological and syntactic information, often involving syllable structure (syllabification tends to be assigned bottom-up as determined by the morphemic distribution of phonemes -- CV always forms a syllable within a morpheme, C+V or C#V does not always). The underlying premise is that surface phones can be mapped to a subset of such sounds, the phonemes, by eliminating all purely "predictable" sounds. But the fact of there being phonetic differences such as "a name" vs "an aim", or the night-rate / Nye-trait pair, would suggest that English has many more "phonemes" than we want to believe (for example, aspirated and unreleased t would be phonemes, because they both occur in the same surface environment). By positing an entity "juncture" which can be positioned just like any other phoneme, aspiration as a contrastive property of phonemes can be eliminated (and there are a lot of other problems that juncture eliminates).
The problem could also be solved by directly referring to morphological or syntactic structure, but at the time (prior to the 60's) this would run afoul of the "single level" theory that prevailed, where phonemic analysis can only consider phonemes, morphophonemic analysis can only consider morphophonemes, morphological analysis can only consider morphemes. There was nothing in the theory that would prevent having "juncture" be the realization of morpheme concatenation, so by translating morphological structure into analogs on the phonemic level, mixing of analytic levels was avoided. This is essentially the driving force behind later (1980's) prosodic theories of the syntax-phonology interface, where rather than letting phonology directly see syntactic structure in determining whether a given rule applies, a phonological analog was constructed (such as "the prosodic word" or "the prosodic phrase"): that way, phonology only refers to phonological things, which are constructed by reference to syntactic things (and similar moves were made for indirectly encoding morphological information). A classic paper using the concept "juncture" is W. Moulton (Language 1947) "Juncture in Modern Standard German"; see also Hockett's 1958 A Course in Modern Linguistics, Lehiste 1960 An Acoustic-Phonetic Study of Internal Open Juncture, and the not totally unrelated 1956 paper by Chomsky, Halle and Lukoff "On Accent and Juncture in English" (not structuralist, though). Scheer A Guide to Morphosyntax-Phonology Interface Theories provides a reasonable overview. Juncture too on a second and more obscure interpretation, as a means of indicating facts of intonation, supplanted by current theories of boundary tones and different kinds of accents. Basically, the reason why you can't find web resources is that the concept has fallen into desuetude in linguistics over the past half century.
