In my experience, I have had an Osram HQI-T 250w/D metal halide lamp fire up and run reliably on a North American 250w S50 high pressure sodium CWA ballast without any problems using the original ignitor on the ballast.
But the point is, the pulse voltage should be sufficient even when components with their tolerances combined in the most adverse combination, include the most adverse operating conditions (mains voltage, ambient temperature,...) meet up. A single lamp in a one single fixture does not prove anything, except that it just /may sometimes/ work. Even when intentionally making the components on the edge of their tolerances (a common practice when testing a new design intended for mass manufacture) would require quite a few of them in the test group to give anything relevant in the test result.
Building a street installation this way will make half of the lanterns dark when the first winter arrives...