Just saw this video, very very interesting to see:
https://www.youtube.com/watch?v=Med6Rkl3eNA

At and after the moment of burst, the existing hot spot is still on the electrodes inside the arctube, not on the external conductors and frame. So the cloud of thermal emission of electrons and consequently path of least Vdrop is between the electrodes in the arctube, and the arc keeps burning there

The pressure in the arctube must have been extreme, if even after the gas volume expands to the whole outer (and pressure drops accordingly) it still keeps a high intensity discharge. This probably would not happen if running on the correct gear

I found it very cool that the pressure was allowed to decrease so much, but the electrodes were still kept hot enough to function as they normally would. It would be cool to see this started again and see the discharge that would likely occur between the support structures

The actual lamp power may be different from the rating. See if you can find the real lamp power, or if it lines up better if you assume something like 39W instead of 35W for the lamp. (This still have to be verified, but lets see if this would explain anything)

The formula used is correct for Xbal, not for Zbal. They are not too much far away from each other, this alone would not likely cause as high differences as you see. Unless you have a formula that can account for Xbal and Rbal separately, use Xbal and disregard Rbal