Your wavelength error trend using just the lamp lines seems similar to mine qualitatively eg post #37. Your variable offset as shown at H alpha is different to what I see though. My calibration is consistently good at H alpha.
I now have some data for the Filly lamp (I used the “cross” lamp which is just Argon) There does appear to be something subtly different. I will report back when I have figured out what is going on.