What an authour expects the alphabet to do, and what others expect that his use of the alphabet was doing may be at odds.
To offer a context for reading with variant spelling in a manuscript, perhaps we as readers of a text without consistency of spelling, like say the
Devonshire manuscript, (or any other text with non-standard spelling) might be able share the experience with a degree of simititude to period readers of the New Testament with variant spelling 5 or 6 centuries after it was the current language. The poem is transcribed (by a modern editor) as it was originally written, but it is quite easy to standardise the spelling (and interpret words whose meanings have moved on) as well.
8 Iff other thowght in me do growe /
9 butt styll to love yow stedefastly /
10 if that the profe do nott well showe /
11 that I am yowrs Assueredly /
12 lett euery welth turne me to woe
If other thought in me do grow,
except still to love you steadfastly,
if that the proof do not well show
that I am yours assuredly,
let every outcome of labour turn (for) me to woe.
If for example, we were to say that Sir Thomas Wyatt (possibly ascribed authour) had a different number of phonemes in his English - the number represented by variation in his spelling, we might ask whether his understanding of the science (or art) of alphabet design was as refined as or own is. Was his variation between, for example, single and double "f" a) significant and if it was, then b) was it phonetic or phonemic. In other words, we are used to a certain set of significances in the alphabet, use our expectation to interpret what is there.
There is a general consensus that the relationship between graphemes and phonemes is not one-to-one. At the beginning of alphabetical writing, the alphabet adopted hadba certain (inadequate) number of symbols, which led to the the digraph innovation - diphthongs and double consonants. Based on the morphology of words using it (π.χ. κράζω), it is possible that zeta was initially used to represent a number of sounds. In other cases the /z/ is represented by sigma too, as in κόσμος.
To balance the argument that a long /i:/ is preserved in Luke, we would have to insist that words with compensatory lengthening for the loss of a consonant such as the ubiquitous ἀποκριθείς were also evidence of the existence of long /i:/ in the Koine period. Point being, that graphemic evidence for an historical loss of a consonant is not necessarily significant evidence for the inclusion of and extra phoneme in a later period of the language.