Just to play devils advocate, regarding the fish plates.... (I have been in the 4x4 community for a long time, I know fish plates are the status quo)
I design bridge cranes and we splice beams together when the span is over 65'-ish. Our beams are A992, not heat treated.
Side note: Can anyone confirm these frames are
actually heat treated? I know a lot of trucks are, just wondered if it was assumed or known on the LMTV. I digress....
We do a full penetration weld on the upper and lower flanges and that takes all of the
bending moment. We do a splice plate on the web, welded on 1 side for simplicity and bolted on the other side. The plate takes all of the
shear load. The surfaces between the web and the plate are not "slip critical". In theory, I want the bolts on the splice to move, they are there for vertical shear only, not bending moment. We don't do any fish plates. This is all structural steel but it is on a bridge crane so there are oscillating and very large loads. Maybe there's no correlation to truck frame rails but maybe the failures you're seeing with small/no fish plates is just poor design. *shrug*
EDIT: Forgot to point out the access holes in the corners of the web-flange are left open. Weld those shut if you want to see stress concentrations and possibly failure.
All that being said, if it were me and I was splicing the frame rails, with a lot of hesitation, it would be 0% welded, 100% bolted.