To buy the argument that a wide Enter key was prevented by keyboard density is to accept that every single keyboard function as implemented is absolutely essential -- that there was no way to combine a couple of menus under a single name, or move a couple of functions to the appropriate menus (where they're probably already available anyway).
Many of the choices for keyboard accessibility no doubt were made due to measured or perceived frequency of use, and higher priority on a large Enter key would have merely forced a few more moderately difficult choices.
So, keyboard density alone wasn't the reason for a small Enter. As mentioned by others, a desire to better fit market expectations was more likely the determining factor.
Hey, if it keeps H-P manufacturing decent calculators, it's a small price to pay.
------------------------------------------------------
By the way, during the Powell's lunch recently, Wlodek Mier-Jedrzejowicz suggested a "modal shift key" (I don't remember whether he used that exact terminology) might have been used, with one push offering, say left-shift, a second push offering right-shift, and a third push clearing the shift. Such an approach would have freed up one complete keyspace (its primary legend, alpha character and both shifted functions), for the expense of making every right-shifted function require an extra keypress.
Another approach would have been to squeeze the cursor control buttons to the left, and put a two-place vertical Enter above the backspace key.
But that's all moot, as far as the current models is concerned. Perhaps a future version (49G+IIxUltraClassic) could have a double-height Enter stuck somewhere . . .