They already had the spatha in the Principate they just kept it for cavalry use where it was tactically more useful. Despite some revisionist scholarship on the matter I still think the later legions that adapted the spatha were far less disciplined and effective in pitched battles than the ones from the empire's golden age. Perhaps part of it was decaying discipline and organisation, perhaps part of it was the later Empire's greater focus on avoiding battle and achieving victory through means such as diplomacy, ambuscade, skirmishing and starvation of the enemy, but the bottom line remains. I see the spatha as a consequence of the change from an army that seeks out the enemy in pitched battles to one that expects to mostly fight smaller skirmishes where a more unwieldy but powerful weapon is to be preferred.

As for the stat-balancing in EB, part of the problem is the R:TW engine. Lethality simply is far better than attack. Smiths and experience can raise attack and defence values very high even if they were low to begin with, whilst nothing changes lethality. Looking through the "Surprisingly good/bad units" thread this theme is apparent again and again. Units with high lethality are surprisingly good, with low lethality surprisingly bad.

I know there won't be any more big changes to EB1, but in theory I think there are two solutions to this: First, smaller differences in lethality. A longsword being more than twice as effective as a non-gladius short-sword makes the difference too pronounced. And second: doing what mods like Fourth Age Total War do, and having all units start out with a lot of experience based on class. If stats are balanced for all elite units starting out with silver chevrons to reflect their elite status, their getting experience won't unbalance the game as they gain it much more slowly. And if it becomes hard to raise attack values through such means, then having a higher base attack actually becomes an advantage that might compensate for lower lethality.