Most extruders (if not all) can extruder faster then the max. printer speed. Letting the extruder "follow" the intake need of the 3d printer thus including acceleration, retract and stopping will make it impossible to get consistent filament width. So feeding directly into the 3 printer without any spool winding/unwinding mechanism that allows for a buffer of filament is (IMHO) impossible. If your goals is to stop using filament but use pellets then you need to find a way to build a hotend that uses pellets as it's source. In other industries those systems already exist but cost an arm and leg Like pressurized hotends feed by plastic from a meltpot reservoir.
You're also talking about ABS but pelletized from recycled Auto Bumpers. Some issues there are contamination and inconsistent pellet size. Contamination can be dealt with by using a meltfilter and mixing virgin ABS in large quantities. I think I read somewhere 75% virgin against 25% recycled.
In terms of "an easy to put together kit " I don't think that (at this moment) the filastruder can be beat I'm not having shares in that company or even a filastruder but that's my opinion based on reading all that's available on it's forum.
I have a Filistruder and it is a great machine. But it is rather limited. ALL of the current machines are. Hugh Lyman has the right idea but no one is jumping onboard.
On a proper extruder line you have to maintain tension. If you do not you get inconsistent dimensional control. Just letting it squirt from the die is poor practice. You need an independent pull drive tugging the filament as it exits the die. Ideally the workflow would be melt, extrude from the die, flash cool, and pull... with level-winding an option at the end.
By doing it this way you can virtually eliminate ovality issues, density issues and all the other common extrusion defects caused by just squirting and praying.
The other thing is machines need are multiple heating zones. Real extruders even tiny ones use 2-3 zones for heat control.
And lastly a proper extruder screw is not simply an auger, it's a complex geometry that melts the plastic, mixes and kneads the melt in concert with the heating zones. Currently we auger pellets into a hot chamber and push on them until they melt. Plastic industry did away with that in the late 1940s early 1950s. It's obsolete for a reason -- It sucks.
This is what we are up against to develop the proper filament extruders we deserve. The technology exists, but I lack the gumption or the technical skill to make such a beast. I'm simply a process engineer, hardware is a different pay-grade.