-3dB isn't an accurate reading for power calculations - it's about half the true value.
Doesn't matter. It's the accepted standard by which we compare lo-pass bandwidths.
Sure - in oscilloscopes.
What ARE you quibbling about? Anybody who knows their way around electronics ought to understand what someone means when they describe the -3dB cutoff of an instrument relative to some centre band frequency. The -3dB point on a bode plot is for any circuit or device, it's got nothing especially to do with oscilloscopes. The whole -3dB thing is so fundamental and universal that it's baked into basic formulae such as f = 1/2piRC.
In multimeters we want the actual RMS value.
Firstly dB are ratios, whether that is in Watts, Amps or, as in this case, rms Volts. Secondly, what is an actual Volts rms value going to tell you that "-3 dB" doesn't tell you faster? It's a fair sight easier to understand a ratio as 3 dB than be presented with two numbers and have to work out for yourself if and whether they are in the ratio 1/sqrt(2).