I did a lot of work on this in the 1980's for a very expensive high end Department of Energy contract with view ports in the suction of a pump, pitot probes, etc.
Our task was to simulate 10 diameters of upstream piping after an elbow. The elbow had turning vanes and 3/16 cell-sized honey-comb flow straighteners, 2 inches thick, on the inlet and outlet of the elbow. Traversing pitot probes were used to get the velocity profile.
The profile was kind of OK, but didn't simulate 10 diameters. And what was just described seems way way fancier than a suction diffuser.
Even so, unless you're on the ragged edge of NPSH performance for the pump, there's a good chance it would work fine with even the elbow (I'm not suggesting that you do this straightaway). If the diffuser makes the situation better for a reasonable cost, even better. Conical reducers with an included angle of 7 degrees or less on the downstream side would tend to stabilize flow, as does small pipes tapping off a large plenum.
You can do CFD if you wish. But the problem with CFD is, when you get a flow pattern, how will you assess it? What is the pass/fail criteria?
It also depends on how much money you want to spend.