I'd like to work with a developer of Arducopter software

Over the last 4 years I have been making no progress. In fact every new copter and every new software upgrade makes my copters perform worse and worse.
It may be that the new software becomes more and more complex with more parameters that need to be carefully addressed. Since I’m not an expert in Arducopter software, before I throw a towel, I’d like to work with a developer who is working on the single-copter software. Perhaps it would benefit both of us as I can provide some valuable data.

Below is a short history along with log files from the latest flights.

I achieved nice and stable flights with Arducopter v3.4 Almost rock solid hover over a point and nice, slow and straight landings. That was 4 years ago and a video of my first copter in Loiter and RTL modes is below. There was quite a breeze. Unfortunately no log:

The latest tests from today are below. It’s the latest copter and software (v4.2). I did it in my backyard to protect the copter from any wind. All I wanted was to get to 2 m, hover over one point and land. All loiter and land or RTL.
As you can see it will not hold position or attitude. Every switching to Land or RTL results with weird behavior. It usually descends much faster than years ago although the land speed is the same (30 cm/s). It will jump up before descending, go sideways, and the attitude log is very much a jig saw.
I will switch to v3.6 to compare, but before I do that I wanted to post what I got to date.

Quick Loiter and Land modes video: https://youtu.be/x_1M-09A97k
Log: https://drive.google.com/file/d/1VJQI4JPpQ1uPNyN6LpEgty_FmW3cc5fW/view?usp=sharing

Attempt to hover over a point; it dropped down by itself: https://youtu.be/UmH34v5OikA
Attempt to hover and quick RTL: https://youtu.be/-NArmxi5i7o
Combined log for both: https://drive.google.com/file/d/1K_SwmAzQqHRG0GuU6hZ9rgcSDG-ssOhv/view?usp=sharing

Bit unrelated but, is it only the GPS modes that are broken ?

I’m saying this because I also struggled with my GPS (though there was a major bug on mine so I’m not sure). Later when I configured a quad with the same components, the pitch-roll axis I used for the single copter were reversed compared to what I was now using for the quad (and the quad GPS modes now worked fine). Commands on the radio controller were doing exactly the opposite movement as to what was happening on the single when in manual and AltHold.

I am still very confused about that and could never troubleshoot it. Maybe I fucked up the setup of my single, and you somehow did the same (maybe there is a problem in the documentation and or the code that makes the roll-pitch reversed).

I want to add that on my single, the attitude displayed in missionplanner was right, North was North, pitching down (arrow on the computer pointed down) was down, right was right. The flaps were acting accordingly (I mean, it was flying if that’s not proof). Same thing on that quad, everything was the same, except the stick of the radio.

(also triple checked the radio config at the time, no major changes on there)

All of this confused rambling by myself to say that maybe, there’s a strange thing on what pitch-roll means and the GPS is confused. If you have a quad on hand you could compare.

I can’t tell because I am not able to fly single copters in Stable mode.
My next move will be to start swapping hardware piece by piece. Compass, autopilot, etc.
I don’t think however it’s the hardware.

Before I do that, I’ll load an old stable software such as 3.6