I realize this is an old post, but perhaps someone will look at it at the present.
I know of one particular failure mode associated with removing a lamp from an electronic ballast. If the lamp is removed WHEN approximately the rated current is flowing through the lamp, a voltage spike (think inductive kick-back) can occur at the secondary of the transformer transferring energy to the lamp. This voltage spike can be reflected to the primary of the transformer and it can take the switching transistors. A common way to handle this failure mode is to use a transient voltage suppressor (TVS). If the lamp is dead, and the ballast can no longer power the lamp, the high voltage at the secondary of the transformer is typically not an issue (thanks to UL).
If you have a magnetic ballast and you do not have a circuit interrupting lamp holder, a great arc can occur and the metal of the lamp holder may not be able to withstand the heat from the arc (this is from a more senior engineer standing beside me as I type).
Of course, if the ballast is attempting to power the lamp, the lamp sockets can pose a safety hazard.