Peter Dohm wrote:
I think that the clutch springs may also be a way.......

Classic homebuilder's disease.

Why adapt a component (with unknown properties, clunky packaging,
and a least one severe drawback) when you can purchase products
engineered for the task? Consider the Goetz, Lord, or Lovejoy soft
couplers. Smaller, lighter, and (the really important part), they come
with complete engineering information.

A quick glance at the Centaflex application data shows 16 different
physical sizes with 34 different torsional spring rates, all with
complete data including nominal torque, max torque, allowable
misalignments, nominal and max twist in degrees, weight and mass moment
of inertia. You also get multiple mounting methods.

If nothing else, it is rather nice to be able to model for
predicted results, then select the required torsional stiffness from a
list.

The greater question, by my way of reasoning, would be how much
additional mass (flywheel)......

You really want to say "inertia", not mass. There are two good
reasons. First, using accepted, correct terms greatly improves
discussion. The subject is complex enough without everyone inventing
or misusing terms on the fly, and drilling yourself on correct terms
will help you with correct thinking. Second, as a practical matter, it
possible to build two flywheels of equal mass moment of inertia, but
unequal mass.

or damping (viscous/hydraulic disk) on the PSRU side of the
clutch/spring assembly would provide enough additional smoothing to
justify the added weight.

In the case of the Suzuki drive, the experimental damper added about
5 lbs and lowered steady-throttle resonant vibratory torque from 180
ft-lbs to 115 ft-lbs. Some portion of that 5 lbs also served as
flywheel inertia, a nice design bonus. A torsional damper does not
operate "on the PSRU side of the spring". It operates in parallel with
the spring, or to be more precise, in parallel with a connecting
torsional stiffness.

Dan