Boy, I guess I'm behind the times...
I used to use Batteries as much like a filter capacitor - I put the battery in past the rectifier - but before the regulation section.
That way, the older style Battery charger method worked as it made the battery able to charge up, but as it obtained more charge - the rectifiers took on the task of recharging to more of a low-level sustaining because of the battery now acting more like a capacitor than a resistive load.
The concept here was to use a smaller transformer like those at R/Shack that had the 18V CT winds - gave you easily 19~25VDC but only 2.5 Amps output at the most - it worked well for loading - because the secondary just didn't have a lot of amps to push into the battery+Supply. I did this so I can keep overcharge rates low so the batteries didn't waste a lot of gassing - so You'd let the battery charge and as it's "float level" rose - the secondary power would also rise and you'd start to see ripple and higher output from the secondary - but the batteries own loading across the bridge rectifier - kept the secondary under enough load to keep the ripple voltage low enough not to exceed the linear power supply input (usually 35 to 36 volts I used LM337 and 78XX series for an older computer drive tower for SCSI drives).
I used two batteries or three if I need -5VDC to provide TTL drive - the 12V simple car battery and a couple of smaller 6V Lawnmower types with "ground" of two (one 6V + one 12V =18V in series) ties to Positive of separate 6V and it's negative to provide the linear negative logic level TTL - the bane was having to use separate wind of another simple secondary to apply the charger to the negative supply - ran some ripple noise issues and then came across an old Tripp-Lite type that had separate power supply secondary winds, one used for their logic board and another for the 1218V side - upon experimentation easily output over 36 volts - and my world smoothed out from there once I used a single "iron" and got all my secondary's tied in proper phase to make the unit charge 18VDC (the two battery tied side - used a modified 5V and 12V 78XX series and I used the LM337 to charge the negative side.
I set it up as more of a stop gap to keep the SCSI drives working on a separate feed - for the intent was to make up a battery supply using a solar charger or portable generator to offset the power needs of older 386 - AT/XT style systems as more of a remote operation "standalone" - which became moot once Laptops became the rage...
So I can see where someone can get the notion that it is possible - simply based on the ideas of how a simple battery charger works, but the complexity rises rapidly once you try to figure out how much resistance a battery shows to the system as a resistive load across it's terminals so the transformer you use is effective yet not there to spike the equipment one the battery is charged up.
I do find it interesting the concept of the battery packs used in todays laptops are of similar design of cell packing and "tap points" to obtain the voltages and currents necessary for those power levels - only now in lithium and Ni-Cad/Ni-MH
And that can also tell you it was a long time ago...