The 2Vrms quasi-standard for consumer digital sources like CD seems to come from the maximum level a traditional -10dBV consumer input should always be able to handle. This -10dBV level is a "nominal" working level (with a slow averaging VU meter indicating "0dB"), not the clipping level, and the actual circuitry always has to have quite some headroom, on the order of 5x-10x, the analog sources of the day required this.
Which amounts to 14dB-20dB of headroom, giving a clipping level of +4dBV (1.6Vrms) to +10dBV (3.16Vrms). 2Vrms is 16dB of headroom, and seems to match the dynamic range (DR) figures of early CDs. Looks like the 2Vrms 0dBFS output was wisely chosen by the Sony/Philips engineers..