Thanks everyone; I *finally* think I am getting somewhere with this drive.
Here is the roadmap of what I've done to get here.
Thankfully the good drive has been good and I've not had to mess with it.
This morning I decided to start fresh and see what I could do with the alignment. I set out to make my own "one track on each side" alignment disk. I used a radio shack bulk eraser to erase a disk in the usual pull it away in a circular pattern while enabled technique I've seen people use on TV's. Sure enough I put the blanked disk into the good drive with the scope hooked up to TP1 and there is no signal when I tried to read a track. This is where I learned the bios will try reading, then do a home or something on the drive to return to track 0 and then repeat the read. It does this 3 times before returning a failure to read.
Then I used the Kaypro copy utility to write a single track to the blanked disk. Well two tracks. I wrote track 40 (which is one sides track 20) and I write track 41 (which is the other sides track 20). Now I have a disk with one formatted track on each side and all other tracks unwritten. The copy utility only gives the option to format a single track on drive B:, so I had to change the drive jumper while the system was up and running with the copy utility. Now I have my alignment disk (at least one that is aligned to my good drive).
I then put in place the bad drive which was somewhat aligned, but not perfect. It was good enough to boot CP/M and load my disktool which I will attach. My disktool doesn't have a lot of options, but one of them is a read track function where you specify the drive, the track, and the number of retries. The command for this is X04099 which is disk 0, track 40, retry 99 times. Now with the scope hooked up to TP1, I can see the amplitude of my single track on the scope. I loosened all 3 bolts and started adjusting. What I learned is that when I tighten the bottom two, it will change it slightly so you need to account for that. My technique was, turn it completely clockwise. Then approach the signal and learn where its maximum amplitude is. Then go back clockwise again and reapproach that position. My scope was set to 200mV per div and if you approach it by don't get to the optimum signal by maybe 1/5 or 2/5 graduations, that will be ideal because as you begin to tighten the bottom two bolts back and forth nice and even, they will pull it into its maximum signal that you originally found. Finally tighten the top bolt carefully as well.
My test for alignment is this. Use Kaypro copy command again. O-X formats A:, and O-F formats B:. Put a disk in each drive and format them both. Then do an O-V to verify the disks are the same. It should pass with no errors. Then swap the disks. DO another O-V. If it passes that then each drive is reading the one that the other formatted. Do this multiple times. I've run an O-V again and again swapping the disks again and again and it is always passing now!
In the end I think this drive was so tough to fix because it had multiple issues going on.
#1 - It has bad tantalum capacitors. The most meaningful one to change was C1 which is used for an opamp that probably amplifies the signal coming from the head. This was the cause of the signal strength changing on a whim and why it would sometimes seem to work and other times fail.
#2 - It has a problem with the 5V trace from the connector to the first parts it connects to. It would get weak and introduce 25+ ohms of resistance choking off the drive for power. I tried resoldering it, but I don't think it was the connector, but the trace itself somehow. I finally got sick of messing with it and ran a stranded wired jumper from the topside of the connector to the capacitor it first connects to. Finally a sub 1-ohm reading of continuity.
#3 - bad alignment. It may not have started with this issue, but it could have been introduced by me trying to previously fix the drive not knowing about issues #1 and #2. My technique above is how I fixed that.
The last remaining thing is that sometimes it seems to start out not knowing what track it is on and then seems to do a home to figure that out. I suspect that perhaps a track 0 sensor adjustment can be done to improve this. After reading the manual I learned that this sensor does not just identify track 0, but actually when the carriage is close to track 0 getting activated somewhere between 0 and 3. Then it relies on the phase of the stepper being at 0 to know it is on track 0. Ideally I would think then you would want the track 0 sensor to activate on the edge tracks 2 and tracks 1. I still need to check that on both drives and see how it performs.