Second Part
m. 19: I would advise against an "8" here, as the melody features the third, not the octave. "" - "6" - ""
m. 20: There is an unprepared seventh there: <c> in the melody against <d> in the bass. And it doesn't resolve down, but up to <d>, which itself is an unprepared dissonance against <a> in the bass. I would recommend changing the bass line to <a> - <e> - <d> or a variant thereof, which would be ii - IV6 - V7 in G Major. This would need "" - "6" - "#", with the <c> over <d> explained away as a retardatio, i.e. a note that should move a second up but does it too late. Your original bass line could be seen as a special expressive device - baroque composers in fact loved dissonances as a "rhetorical figure" -, but it would be more in place in a mournful, sad piece, not in a rather pleasant sonata.
m. 21: If you want, you could make the cadence stronger by replacing the half cadence on d with a major cadence, by changing the bass line and writing the quavers <h> and <c#> instead of the crotchet <g>. However, it would destroy the rhyme with the bass line of m. 19. So, just a suggestion.
m. 22 : "#" - "-" - "6" / "6"
m. 22 to m. 23: The resolution of the Seventh in m. 22 by a jump to in an open fifth with the bass in parallel motion is rather harsh, even if the discant jump is just an octave transposition. One solution: Shorten the <d> in measure 22 to a quaver, let it precede by rest, transpose it up an octave. Then: "" - "#" - "7" . Second, even better solution: Use the parallelism to m. 24/25. Thus, end m. 22 with the quavers <h c> in the discant and a <g> in the bass, to which the <f#> is pointing anyways. Figuring: a simple "".
m. 23: "" - "#" - "7 #", as you need the seventh for the <c> in the discant. If you want to give it a stronger cadential feeling, like in a Monte, you could shange the bass line to end on <f#> (figuring: "6 5") and have a <g> (figuring: "") in m. 24, and the following two bass notes shifted accordingly. (#6 is normally written as a "6" with a slash through the arc, but I cannot type it.)
m. 24: I guess you wanted to include the melodic <c> in the bass and therefore opted for a "6 4" on the bass <g>. However, this would imply <c e> in the right hand, which is fine with an <a> in the bass. So, I would propose to change the bass note to <a> and sharpen the <c>, giving the feel of A major. Thus, the figuring with my proposed changes would be "" - "6" - "#" for a bass line of <g a h a>, with the first two notes being quavers, the rest crotchets. For your original bass line, I would prefer "6" - "6 4" - "" and treat the <c> as passing dissonance. This also helps with fitting in your proposed "6 4". However, for more harmonic variety, a "6" - "6" - "" would fit as well.
m. 25: The bass here should be a transposition of the one in m. 23, like the discant. Thus, you either have <e d> as the last two quavers, or something ending on <g#> if you changed the bass line in m. 23 to end on <f#>. If its <e d>, the figuring would be "#" - "5# 3#" - "7 #". The <es> in the discant should be reinterpreted as <d#>. Your original bass would suggest "#" - "5# 3#" - "#", so just like you have written, except for a "#" instead of "6 4".
m. 26: Again, I would write an <a> in the bass to strengthen the cadential effect, so the quavers <a h c# d> would result in "" - "6" as figuring.
m. 27: Just erase the "4", the rest is fine.
m. 28: Just replace "8" with "7".
m. 29: "" - "6" - "" , so just shift the "6" one quaver to the left.
m. 30: Again, you go in parallel from a fifth into a tritone, which is no problem in itself, just for the voice-leading of the inner voices, especially to the following sixth chord. You probably aimed for something like a motivo di cadenza. I would suggest another bass line, outlining: <g> - <f> - <e> (which can be embellished), and the figuring "" - "6" - "6 5". Retaining your bass, I would suggest "" - "" - "6 4 2", so just "" instead of "8".
m. 32: "" / "6 5"- "" / "9" - "8" / "7" - I think following the melodic descent could be a good choice. The chords mainly move in crotchets with a discant voice in quavers.
m. 34: "6" - "6" - "" or "6" - "6" / "" - "". I would work out the voices of this measure to decide whether a "7" on the last <g> is a good option.
I probably introduced some mistakes of my own making, so please don’t hesitate to ask any questions or for clarifications, @Maarten Bauer.