Q. final [t] became [s]; [ts] > [ss]
In the Outline of Phonology [OP2] from the 1950s, Tolkien mentioned a late Tarquesta [TQ] sound change of final -t > -s driven by euphony:
In TQ final -t was often lost by grammatical not phonetic change. But where retained, either as an inflexion or as a reduced final in suffixes or compounded stems, it eventually became s. This probably originated in a dissimilative tendency, since this s is earliest found at the end of final syllables beginning with d, t. Sequences such as ondot were apparently not felt to be euphonious, and replaced by ondos &c. But later a weakening or opening of final t > þ, s seems to have become general. Similar developments (but not identical) occurred in Vanyarin. Here t remained, but later > þ, and it was only then that dissimilation set in: in sequence þ·vowel·þ > þ·vowel·s. This s then became general (OP2: PE19/104).
This sound change was not mentioned in Outline of Phonetic Development [OP1] from the 1930s, but the descriptions of final consonant developments in that version of the document were incomplete. The sound change was mentioned in the Early Qenya Phonology of the 1920s, however:
No important changes have affected final consonants since the time of these drastic changes. In Old Qenya began a tendency to spirantalise the one remaining final stop (t, nt). This, being favoured by dissimilation, established itself after vowels immediately preceded by t, nd (so that already in Old Qenya we get ondos for *ondot). In modern Qenya final t is everywhere pronounced -s, except in Inwian; and nt > n (from ns > nz > nn).
Neo-Quenya: For purposes of Neo-Quenya, I think it best to ignore this sound change. Final -t is very common in Quenya, and implementing this sound change would require major revisions to both vocabulary and grammar. Among other things, the dual would need to be revised and the distinction between singular and plural object inflections would be lost. Since we don’t know how Tolkien imagined these grammatical changes playing out, it is better to assume the final -t was restored from spelling after Quenya became a dead language, or that this sound change simply did not occur at all.