Well...now that the thread has been derailed into a discussion of the definition of graupel (which I think is warranted)...
The AMS glossary of meteorology's definition of graupel is
AMS glossary of meteorology said:
Heavily rimed
snow particles, often called
snow pellets; often indistinguishable from very small
soft hail except for the size convention that
hail must have a diameter greater than 5 mm.
Sometimes distinguished by shape into conical, hexagonal, and lump (irregular) graupel.
I think the main tipping point in the distinction between graupel and hail is the size...that 5 mm threshold. Below that, any frozen particle that isn't clearly just a snow or ice crystal could probably be considered graupel, and that does include hydrometeors that began as liquid and subsequently froze. I do think that some degree of riming is necessary to distinguish graupel from sleet at the ground, though. However, most microphysics schemes do not specify a sleet category for hydrometeors, although that is getting at your last point that hydrometeors exist on a continuum of particle descriptors.
Hugh Morrison at NCAR has been at the forefront of this continuum of particle behaviors by designing what is called the predicted particle properties, or P3, microphysics scheme, that has yet to catch on to regular use in the NWP community, but is a step in the right direction towards abandoning the use of discrete precip types and towards greater realism. Instead of assuming the usual 6 or so hydrometeor forms (rain, snow, cloud water, cloud ice, graupel, and hail [which many schemes lump in with graupel]), superparticles (assumed groups of particles) are prognosticated to have various types of particle properties (mass, mass of rime, volume of rime, and number concentration within a unit volume) which can evolve freely. Particle types can subsequently be diagnosed after the fact from the values of those particle properties.
The paper can be found here:
https://doi.org/10.1175/JAS-D-14-0065.1