In number 1 (below), ^5 rises to an accented ^6, which reappears over the tonic two bars later. This is audible as an almost direct resolution because of the distinction in register--but I admit that the gap of two bars, causes me to hesitate a bit. But there is no doubt in the variant at (c), where the chromatic passing motion is expanded and ^5 follows just a bar later, the label "almost direct resolution" is obvious.
The tonic 6/4 as tonic-functioning in its immediate context—as in bars 3-4—is common in Schubert's dances, especially the Ländler, but not in those of other composers.
Note that the accented dissonance effect in bar 1 is repeated every two bars in the first strain. Thus, the 9 in V9 is linked to the milder 6-5 over I (or literally 9-8 over the bass Eb), to the very traditional 4-3 at (b), and even to the chromatic passing tone B-natural5 in bar 7. The flagged high note in bar 6 is a "one-leap-too-far" figure that is a cliché within the dominant of the cadence. (Here, the 4-3 that begins the phrase is overtopped by the leap up to Eb6.) To close the second strain, Schubert repeats the figure—with a dissonance (Db6) and timed to the two-bar groups of the first strain.
One can easily hear the first strain of no. 3 as a variation of no. 2, the famous Trauerwalzer. Another sentence, this first strain has a leap and passing tone figure at bar 1.3, repeated a step higher at bar 3.3, then again in bar 5.3. The second phrase is repeated to close the second strain, at (d).
In n13, an expanded V9 with a direct resolution at (a), repeated in the second phrase. The flexible treatment of accented dissonances that is typical of the waltz throughout the nineteenth century is already apparent here in Schubert. In the second strain, boxed at (b) and later, the dissonances are traditional accented passing tones. But if we regard these as parallel to the opening, then G5 is the dissonance and F5, the ninth of the chord, is the resolution.