I believe you are on the right track looking at reduced voltage to meet your various requirements. (A wye/delta system changes to motor configuration to reduce volts/turn, so can be considered a form of reduced voltage.)
You wish to reduce motor torque and current using an ordinary squirrel cage motor, without using power electronics.
You agree that using a VFD to change drive frequency is the best way to control this motor, and you _are_ using a VFD, but want a non power electronic approach as a backup. You also agree that absent a VFD, various old school techniques such as wound rotor machines would be an option, but now we are looking at much more custom/expensive hardware given present day manufacturing.
So given these design constraints (no VFD, 'modern' COTS motor), voltage reduction is the only tool you have available to reduce torque.
mentions in post 18 that there is a voltage and current spike during the wye/delta transition. I don't know how much of this transient is caused by the phase angle change and how much by the effective voltage change. This spike is presumably a short transient which won't cause appreciable heating, and also won't damage power electronics that isn't present in the system.
Using a motor starting autotransformer https://americas.hammondpowersoluti...ol-automation/motor-starting-autotransformers
and suitable tap selection contactors you can reduce the voltage applied to the motor, without any significant phase change. This would let you drop torque in a non-electronic fashion. Standard tapping is for 50%, 65%, and 80% voltage, giving you a selection of torque values to consider (25%, 42% or 64%)
I personally feel that at the 50 Hp size, spending money on two VFDs is probably the best 'bang for the buck' in terms of redundancy. But maybe you need this system to function after an EMP or something. So I'll needle you a bit but respect the design constraint you are setting