End or top, either way is fine, whichever is easier to build, just as long as you are under the fluid level, and not causing turbulence at the surface. I'd stay a couple inches off the bottom, too much velocity and turbulence if too close. Cut the pipe end at an angle to very slightly help in the diffusion. Not a big deal, but helps somewhat. Also tends to direct the flow towards the cut side.
The air doesn't come from the return in any way. Some air is dissolved in the oil but that stays put. Mostly air is entrained in the fluid and comes in through leaky suction piping, or shaft seals on pump, or turbulence at the surface of the tank fluid. So you want diffusion to slow down the velocity, move the oil slow and evenly through the tank, and allow time for entrained air to move upward towards the top.
Also, ideally, moving along the outside walls of tank helps to dissipate heat better than stagnant fluid at the walls.
Industrial tanks tend to be just big square boxes, because space is more available. Hence the 1x, or 2x or 3x pump flow rules of thumb form the old days. Tanks on mobile equipment can get pretty exotic because space and weight and shape are at a premium, and no way will a huge tank fit. Most hydrostatic drives have tanks less than half of the charge flow. (Or power steering pumps, reservoirs of a pint or so.) That takes more attention and complexity in design to get nice diffusion, holding time, deaeration, settling areas, cooling, and preventing vortexes over the suction port. Do some google searching and you will see some fairly exotic tank designs. ONe research tank, made in plexiglass for visibility, created flow in a spiraling motion inside (without baffles) that moves the fluid like it was being pulled inward to a black hole/death star. Never heard if anyone used the concepts in a production design. Complicated though.
k