The green gradient on the side is actually present in all three channels, but a bit more in green and red:

(this histogram also shows that your blue channel is somewhat undexposed)

So the problem has nothing to do with your methodology, it is part of your data (parasitic light?).

To restore the image from the individual channels, I used:

• Open the first channel image in Gimp.
• Convert the image to RGB (Image > Mode > RGB)
• Start the bucket fill tool
• Set the foreground color to the intended color of the channel
• Set the bucket-fill tool to Multiply mode

• Bucket fill the whole layer. That should set all the white to the channel color (so if you started with red, all your stars will turn red)

For the other two channels:

• Add the channel with File > Open as layers
• If necessary adjust the layer position
• Bucket fill in Multiply mode with the required color as above
• Set the layer to Addition mode.

If you wan to go to the next level:

• For each color you make a group with:

  • the stars image at the bottom
  • a layer filled with the channel color, set to "Multiply" mode

• You set the top two groups to Addition mode

This way you can do fake colors by re-filling the color layers with other colors, and if necessary play with the opacity of the two top groups (if you want to play with the opacity of the bottom thrid group, add a black background layer outside of the three groups).