I have done it in the past. Cut at about 20deg so you have a long slanting cut surface. Harden and then rub on a stone or diamond slip to get a nice crisp edge and feed it into a slightly undersized hole at slow speed or turning by hand.

Reamers will produce oversize and tapered holes unless they are allowed the freedom of a floating reamer holder or the tailstock is aligned to better than one tenth of the precision required.

If you make your reamer on the end of a 330mm length of silver steel there will be enough flex in the 3mm dia for it to align itself, floating or not.

Most commonly available reamers being H7 or H8 are a plus size tolerance so I would expect them to be over nominal.

After thinking a bit about this I am attempting to analyse how it works. Lets assume the flat face is upwards in case it aids the description of operation. The flat face width starts small increases to the rod diameter and then decreases, with that peak being at half way up.
As the reamer enters the hole the tip is not cutting just guiding and as the hole is undersize the reamer is being pushed upwards. It starts to cut as the width of the flat face matches the undersize hole diameter not at its max design width -almost- but because it is pushed upwards it is contacting above centreline and therefore narrower width. Also the cutting edge has negative rake.
Even though the hole is being enlarged the tool remains above centreline because the tip is still resting in the pilot hole and therefore the bottom of the tool is in mid air. At some point the full width section has to get into the hole despite the tip still being above centreline.

At this point my imagination is getting a bit strained about what happens. It does however explain why some people suggest boring the start section of a hole to be reamed.