MIMF

I agree with Barry, use a flat leveling beam.
I’ll add that I now use a six inch long leveling beam.
I found that when using a full length beam, the first few frets were always ground more than the last frets because the last frets are closer together - more material.
So far, my results are good.

What Barry said: a flat leveling beam used along the string paths making a conical section fret surface. If the fingerboard was cleanly leveled (a radius beam for that is fine—the differences between the conical section and the cylinder are small), the frets will only barely be touched by the leveling beam before they are perfect height. Not much metal needs to be removed. If you mark the top of each fret with a black marker before stroking with the beam, you will be able to tell when all are touched. When I have problems with a fingerboard that is leveled this way, it is usually because of something else other than the fret geometry. A couple of times, I found a loose fret—down when I dressed it, up just a bit when not being played. One time, I struggled with one string making a buzz at just one fret. I could find nothing wrong. When I loosened the string to replace it, I found a kink just past that fret.

Agree with you that fret heights should be added to the curve of a radius sanding block.

Barry's last paragraph above is really important to understand. You want the frets to be level along the string path. A tapered neck will end up having a very slightly different radius at each fret locations. As you level with the straight beam, keep the string paths in mind and move the beam across the board smoothly. I don't see how this can be done with a fixed radius sanding block. If all the frets are level along the string paths the first fret and the last fret will not have the same radius.

Yes, if you used a fixed radius block to level the frets, you would actually have a slightly humped fret surface at the edges of the fretboard where the e and E string lie, due to the tapered fretboard. Set the action low and you would likely get buzzing on those strings.

Barry, would this be from the radius beam not running perfectly parallel with the centerline of the neck?

Now that I am looking at my last post, I am not so sure it is accurate. But I still stand behind the statement that a radius block is the wrong tool to level frets.

No, your post was correct. If you put a straightedge on angle across your radius beam, you will see a gap. You can’t make a conical surface that will exactly mate with a cylinder. The string paths taper from the bridge to the nut. For the frets to provide a flat surface under the string path, the fret surface has to be conical.

You're right Bob. My brain wasn't working this morning.

I'm with Barry.

There was a massive discussion a few years (decades?) back that ended up in an argument between the "coneheads" (level fretboard/frets to a compound radius) and the "flatheads" (level fretboard/frets flat).

IMHO, as long as the frets are level to the relative distance of the strings (affected by nut and bridge height), you're good.