This is the first I've heard of this issue on this forum.
Most good DMM power supplies are isolated from its measurement circuitry. To illustrate, see the Fluke 87-I regulator section in the schematic linked. Thus, the input section floats relative to the 9v supply. Probing the battery on any DMM with its own probes even with proper polarity should not be done, not just on Flukes.
Note, you can get old DMMs and some modern panel meters that have common ground with it its power supply, such as the negative terminal of a 9V battery.
What to do now? The manual of the 87 can be illustrative but it may not be relevant to the details of the 87V as its 4 versions old.
I'd suggest trying any of the following procedures as you've nothing to lose, on the assumption its not physically damaged, try it one at a time and see what happens:
Remove the 9v battery, and short the 9v inputs assuming waiting 20 min with battery removed was not long enough
On the service manual of the 87V, do all the performance check tests, the LCD, the fuses, etc., see if will reset the battery annunciator. Also, you'd like to know if the DMM is actually truly working right. The hope is that when in 'test' mode the firmware executes some form of full reset before executing the test sequence and could reset the battery reference indicator.
If the above doesn't work, engage the calibration mode, then abort it.
If the above doesn't work, reset the security password for calibration to "1234".
All the above assume that the firmware will begin any test sequences with a full reset, a function that isn't available directly through another user control.
Keep us posted.