Having another 5V for the ACS712 is a good idea but it has to be analog. Not another adapter using using switching.
I also made sketch for connection I have external supply so I will not use second divider
You are missing the point! Could you assure me that your external 5V is +/- 0.01V all the times! If not then you are back to square one.
The reason is that you are using a very small portion of your current sensor and the voltage slope against current difference is very low. The offset of the current sensor at 0A is VCC/2. So if VCC varies the output voltage will and using a constant offset ther is pointless.
So the best approach will be. (Looks like I repeat myself all the time).
1 - Read VCC and ACS712 and get their real Voltage after conversion from the resistor ratio. You need to display the real Voltage at VCC and ACS712.
VCC ~ 5V and ACS712 should be ~2.5V
2 - Correct the ACS712 signal against VCC.
new ACS712 Voltage = Current ACS712Voltage * 5V/ VCC voltage read from ADS1115.
This way the signal offset will be always 2.5V or very near. A simple adjustement will work.
This mean that you need to measure the resistor with a multimeter to get the correct ratio. For the resistor divider.
B.T.W. the 1K resistor is not needed in your schema. Thevenin theorem explain that the 10K and 4k5 do the same thing.
You should have all the calculation in your hand now. Try to figure it out. If you can'y I will see if I could create a python script with all the calculation but i'm buzy today.