At this point, it can get to be a bit of a guessing game, since very small moves in the secondary can make a big difference. Most likely, the secondary mirror is still out of position a little bit, either rotationally or axially. It could, however, be the way the primary mirror is made. It could be off a tiny bit. However, I think secondary mirror position or rotation is probably the culprit. If we call the focuser up and the opposite side from the focuser down, if you have to move the primary mirror up or down a bunch, compared to its position when you are aligning it from the front of the scope, I am thinking it means there is something still wrong with the mirrors position along the axis of the scope. If the movement is in a different direction, rotation of the secondary mirror is most likely involved,
I do think a star test is in order, because that is the final verdict on collimation. You can make adjustments and try to get the primary mirror into its best collimation by judging how a bright star looks through your 9mm eyepiece with the star centered. When you turn a screw, the whole image will move. In order to make a judgment on the move you made, you have to move the scope to get the star back into the center of the field again.
I have a mental block on remembering which way you want the star to move in the field in order to get the defocused image to go circular, when it is not round or you have coma. Seems like you move the star in the direction of the elongated part or the direction the coma is going. However, I always have a doubt that I need to go in the opposite direction.
Strengthening the focuser is a good idea and helps keep it truly perpendicular to the tube and pointing directly at the main tube axis. To get things dead on, the axis of the draw tube and the axis of the main tube have to intersect at a 90 degree angle, which is hard to hit or even to measure.
I have made attempts to verify the focuser position and direction using a laser collimator before, but that gets pretty exhaustive. I am reluctant to suggest that you do that. My method was to pull the secondary mirror out of the way, then use the laser to shine on the far side of the optical tube. Using a ruler, I made a best guess at how far down into the main tube the red spot was located, and then measure the best I could to the center of the draw tube from the front of the scope. The distances should be the same. That is terribly crude. I then eyeballed the position of the red dot circumferentially by peeking around either side the spider leg adjacent to the red dot. Again very crude.
I do not know if the final position was really good enough, but I did find the dot maybe a quarter of an inch off of where I thought it should be and tweaked the metal a bit to get the dot where I thought it should be. After I fixed that finder attachment piece, I checked the position again and had to tweak it back to about the same spot it started at. With that said, I am hoping you do not have to so something like that. You have worked over that finder piece already.
Taking the secondary mirror out, making changes like that, then getting things all back together is a pain. Then you have to go through the whole collimation process again.
Well, that is probably enough advice and rabbit trails followed.
Hope it all helps,