Hello. I believe to have discovered a parsing problem of Warcraft II's XMI. (eg. 419.xmi)
The problem about VLQ2's reading is the forced stop at 4th byte, for a 508 maximum value, whereas my XMI file can admit delta which are longer. This will likely produce a misreading of a series junk events.
In case of Warcraft II, I have end-of-track at delta value 166003.

This is a report based on software which uses WildMIDI's XMI parser for preprocessing sequencer inputs.
Here was the proposed fix and the problem description: Wohlstand/libADLMIDI#158