I fail to understand the difference/challenges between the two situations.
Plugging a unit (or circuit card) to a live backplane.
Vs
Plugging it to a dead bus (or inactive power supply) and then turning ON.
I can imagine a contact debounce/glitches while inserting to a live source. Is that all? I see many hot swap controllers specifically designed for the former case.
Contact bounce as you mention, yes, but also the board being connected will have filtering capacitors on the input that will draw a large current when connected to the backplane. In a server, for example, it's desirable that some maintenance is able to be performed without shutting the system down, hence the need for a device to permit connecting a board to an energized bus. By slowly ramping up the current into the board, a hot swap controller will prevent large inrush currents from flowing which could cause sparking and contact erosion or even disturb the supply bus from overcurrent/undervoltage.
Since some hot swap controllers use a sense resistor to provide current feedback on the ramp-up, it's easy for the manufacturer to add in useful features such as efuse capability, basic telemetry (e.g. current/voltage sensing), and an enable signal to turn on voltage regulators only when their supply capacitors are fully charged.
Apart from current draw surges another problem is that damage may occur if inputs to chips on the inserted board are powered up before the chips themselves are powered up. Typically chips can tolerate only 0.3 to 0.7 V above VCC or below ground.
When the recommended procedure is followed the power will rise on all boards simultaneously and so any signal from one board to another would always be within the power-rail voltages.
