Not sure which 300B thread you are referring to, but my view is that it is reasonable to apply margins like the following for a secondary V-0-V winding being full-wave rectified in to a capacitor filter:
V' = V x 1.1 (for +10% mains ACV) x 1.1 (for a mains supply surge/swell) x fudge factor for difference between transformer rated primary voltage and what is now normally provided (eg. 125/117 = 1.07).
The peak diode PIV requirement is then 2 x sqrt2 x V', given a capacitor input filter, and no rectifier diode drop and that there may be no loading initially.
There is typically an inherent PIV margin when using a generic diode like a 1N4007, where diodes would be screened to ensure they will always meet 1kV, so statistically they manufacture for an actual PIV capability of 20-30% higher. If a single diode doesn't have an acceptable PIV rating, then using series connected diodes requires some PIV derating for each diode, and perhaps use 700V for a 1kV rated diode to cover imbalance. So there can be a lot of built in ss diode PIV margin even before transformer voltage margins are considered.
