In order to address your problem and questions, let me make some declarations first.
The Clock Effect is the statement that: between two timelike-related events, the elapsed proper-time along the inertial segment joining them is longer than the elapsed proper-time along any other timelike-worldline segment joining them. Geometrically speaking, this is the spacetime-analogue of the triangle-inequality.
The Twin Paradox is the attempt to dispute the Clock Effect by analyzing the problem from the viewpoint of the non-inertial worldline with the hope of obtaining the same conclusions as gotten from the viewpoint of the inertial worldline... thus nullifying the effect, resulting in no time dilation.
Let me apologize for the length of this answer.
The twin paradox has a minefield of gotchas.
(And, I think your setup has hit one of the gotchas.)
So, I'm looking to cover many of the bases.
With spacetime diagrams and calculations, I verify your first three paragraphs.
Then, I discuss the rest of your question.
You can, of course, skip down near the end for a discussion of your questions.
For concreteness and arithmetic-simplicity,
let us suppose that the "travelingTwin" has
$v_{outbound}=3/5$ and $v_{inbound}=-3/5$ as viewed in the inertial frame of the "homeTwin".
Further, let $L=3$ be the distance to AlphaCentauri, which is at rest in the hone-twin's frame.
We follow your convention that the turn-around event B on AlphaCentauri is the common origin $(x,t)=(0,0)$ of the travelingTwin and AlphaCentauri.
Let's work out some numbers for convenience:
With $v=\frac{3}{5}$ and $L=3$, we have $\gamma=(1-v^2)^{-1/2}=\left(\frac{5}{4}\right)=1.25$ ,
$\gamma L=\left(\frac{5}{4}\right)3=\left(\frac{15}{4}\right)=3.75$,
$\gamma (L/v) =\frac{\left(\frac{5}{4}\right)3}{\left(\frac{3}{5}\right)}=\frac{25}{4}=6.25$,
$L/v =\frac{3}{\left(\frac{3}{5}\right)}=5$,
$(L/v)/\gamma =\frac{3}{\left(\frac{3}{5}\right)\left(\frac{5}{4}\right)}=4$,
and
$(L/v)/\gamma^2 =\frac{3}{\left(\frac{3}{5}\right)\left(\frac{5}{4}\right)^2}=16/5=3.2$.
I have drawn a spacetime diagram on rotated graph paper so that we can more easily visualize the ticks along worldlines on the spacetime diagram.

So, starting at the separation event,
the ${\rm\color{red}{homeTwin\ ages\ 10\ ticks\ along\ the\ red\ worldline}}$ until the reunion event,
while the ${\rm\color{blue}{travelingTwin\ ages\ 4\ ticks\ along\ the\ blue\ worldline}}$ to
the turn-around event B,
then ages ${\rm\color{green}{another \ 4\ ticks\ along\ the\ green\ worldline}}$ [for a total of 8 ticks] to the reunion event.
When the origin is at the separation event,
this is the standard formulation of the problem.
Here, we have drawn it in the $\rm\color{magenta}{AlphaCentauri\ frame\ (what\ you\ call\ O'\ and\ O'')}$ with the origin at event B.
Note that, in this frame,
the event you will call A is simultaneous with the separation event,
and
the event you will call C is simultaneous with the reunion event.
The AlphaCentauri-(x',t')-coordinates of the various events are shown.
In agreement with your second paragraph,
$(x'_A,t'_A)=(0,-\frac{L}{v})=(0,-5)$ and $t'_B-t'_A=0-\left(-\frac{L}{v}\right)=\frac{L}{v}=5$.
So, now we draw this is the frame of the $\rm\color{blue}{outbound\ frame\ (what\ you\ call\ O)}$. I have included the lines-of-simultaneity for this frame. Later, I will redraw the "standard diagram" with these lines.
The outboundTwin-(x,t)-coordinates of the various events are shown.
In agreement with your second paragraph,
$(x_A,t_A)=(\gamma L ,-\gamma\frac{L}{v})=(3.75,-6.25)$ and $t_B-t_A=0-\left(-\gamma\frac{L}{v}\right)=\gamma\frac{L}{v}=6.25$.

Up to event B, the travelingTwin (O-frame) and AlphaCentauri (O'-frame) are both inertial observers.
However, turning back and continuing onward along the travelingTwin's worldline,
the travelingTwin is no longer inertial because the travelingTwin's worldline now has a kink any inertial reference frame. (No Lorentz boost can straighten that kink.)
Physically, a ball initially at rest on a frictionless table on the travelingTwin's ship will move at event B if the travelingTwin turns back [by firing its rockets].
Since AlphaCentauri (O''-frame) maintains the same velocity (magnitude and direction) of the O'-frame, it continues to be an inertial observer [so, in fact, the O''-frame coincides with the O'-frame].
To be clear,
the $\rm\color{blue}{outbound\ leg\ of\ the\ travelingTwin}\ is\ inertial$,
the $\rm\color{green}{inbound\ leg\ of\ the\ travelingTwin}\ is\ inertial$,
but this piecewise-inertial combination of worldlines is non-inertial.
So, along this $\rm\color{green}{inbound\ leg}$, we draw another spacetime diagram (for convenience---one could learn to read everything from one spacetime diagram [on rotated graph paper]).
This leg and Its coordinate deserve its own set of "primes": O''' with $(x''',t''')$
Because the inbound speed is equal to the outbound speed, some of the coordinate-values are similar. However, if the inbound speed were different, the numbers would be different.. introducing some merely-numerical complication.
Note that the coordinates of an event in this inbound-leg's coordinates
is different from that of the outbound-leg.
$(x_A''',t_A''')=(-3.75,-6.25)$ and $(x_{sep}''',t_{sep}''')=(0,-4)$ and $(x_{reu}''',t_{reu}''')=(-7.5,8.5)$
$(x_A,t_A)=(\phantom{-} 3.75,-6.25)$ and $(x_{sep},t_{sep})=(-7.5,-8.5)$ and $(x_{reu},t_{reu})=(0,4)$
Thus, for this non-inertial observer,
One has to somehow splice the two spacetime diagrams together, effectively devising some rule to make coordinates in one-to-one correspondence with ALL events in spacetime. (No double counting of events and No omission of events.)
(For inertial observer, no special rule isn't needed... another indication that non-inertial observers are distinct from inertial ones.)
In agreement with your third paragraph (with $'''$),
$(x'''_C,t'''_C)=(\gamma L ,\gamma\frac{L}{v})=(3.75,6.25)$
and $t'''_C-t'''_B=\left(\gamma\frac{L}{v}\right)-0=\gamma\frac{L}{v}=6.25$.

Now for your comparisons in your fourth paragraph.
I will insert this hybrid diagram (blending the previous two diagrams)
which has not yet had any rule devised to splice this into a valid spacetime diagram
(as described above).
(If you naively follow
the $\rm\color{blue}{outbound\ leg}$ of the travelingTwin with its lines of simultaneity,
then switch at B to
the $\rm\color{green}{inbound\ leg}$ of the travelingTwin with its lines of simultaneity,
then splice the lower half of the outbound diagram
with the upper half of the inbound diagram
,
note that you have omitted the midpoint (tick 5) of the homeTwin's worldline on your spliced diagram.)

With the travelingTwin's noninertial hybrid $O-O'''$-frame,
the hybrid quantity $(t'''_C-t'''_B)-(t_B-t_A)=2\frac{\gamma L}{v}=6.25+6.25=12.5$.
Because of symmetry in the problem, $t'''_C=t_C$ and $t'''_B=t_B=0$,
so
$t_C-t_A=2\frac{\gamma L}{v}=6.25+6.25=12.5$.
With the AlphaCentauri inertial $O'-O''$-frame,
the quantity $(t''_C-t''_B)+(t'_B-t'_A)=\frac{L}{v}+\frac{L}{v}=5+5=10$.
And, finally, to your questions.
Did Alpha Centauri age less than the twin between events A and C, while the twin aged less than Alpha Centauri (and the twin who remained on Earth) between the events of departure and arrival of the twin in the original formulation of the problem?
Here are various time-intervals.
I'll let you do the comparisons.
Along AlphaCentauri's inertial worldline ABC,
we have that AlphaCentauri aged (from the previous part),
$(t''_C-t''_B)+(t'_B-t'_A)=\frac{L}{v}+\frac{L}{v}=5+5=10$.
For travelingTwin's noninertial worldline, events A and C are not its worldline.
So, we use the travelingTwin's lines of simultaneity to assign times to events A and C.
We have (from the previous part),
$(t'''_C-t'''_B)-(t_B-t_A)=2\frac{\gamma L}{v}=6.25+6.25=12.5$.
(What would this mean?)
Along the homeTwin's inertial worldline sep-reunion,
we have that homeTwin aged
$(t^{home}_{reu}-t^{home}_{sep})=10$.
Along the travelingTwin's noninertial worldline sep-B-reunion,
we have that travelingTwin aged
$(t'''_{reu}-t'''_B)-(t_B-t_{sep})=2(L/v)/\gamma=4+4=8$.
For AlphaCentauri's inertial worldline, events sep and reunion are not its worldline.
So, we use AlphaCentauri's lines of simultaneity (which coincide with homeTwin's lines of simultaneity) to assign times to events sep and reunion.
We have,
$(t'_{reu}-t'_{sep})=\frac{L}{v}+\frac{L}{v}=5+5=10$.
(What would this mean?)
Is this a valid method to debunk the twin paradox?
NO.
As stated in the beginning of my answer,
the Clock Effect and Twin Paradox compare elapsed times
along two worldlines with the same endpoint events... the spacetime-version of the triangle-inequality.
So, the comparison is between:
homeTwin's 10 ticks from sep-reunion
and
travelingTwin's 8 ticks from sep-B-reunion.
These describe the legs of a triangle.
Events A, B, and C do not form a triangle for the trip of the travelingTwin.
And the noninertial travelingTwin spacetime-diagrams and calculations of the time-intervals between events A and C it did not visit (to get 12.5) is suspect.