Very rarely there are two UPDATE_SHAPESHIFT_FORM events while turning back into normal.

When this happens, the second event is probably the one that should start a timer to update the shoulder offset, because then the next UNIT_AURA is too early.

To fix this, UPDATE_SHAPESHIFT_FORM could have to always stop all currently queued timers (cosFix_waitTable = {}) and start a new one, instead of setting activateNextUnitAura = true.