The rarity of a feature is more than enough to fulfill this goal.
And - you are exactly right. Some decks truly are vague and their flexibility is meant to provide for randomness and, hopefully, replayability.
I can go into details with you on what exactly the game does and why each deck is out there - but that might take too much of your time. (it takes me a while to explain, while the initial testing demonstrated that it’s quicker to throw the players right into the game, rather than explain everything in advance).
You are exactly right. 
The loot deck has its tiers. This affords for the desired randomness. For instance, each tier has one card whose description is: draw a loot card tier +1. Which means that a powerful item, entire riches can be accidentally stumbled upon, but this is very, very improbable.
You are very much right. The downtime waste is a very high risk with this game design. One of the goals of having this simulation was figuring out how to reduce this and combat the problem by any means.
You are very much right: separation of decks is a plausible solution. The idea is to have the working version set up, so that the optimum is quicker to be found. For instance, the varied terrain encounters in a single deck, in a way, reduce the need for duplicates and can achieve several goals at once (introduce rarity of a certain encounter and its rarity different based on the terrain type, which would motivate the players to avoid or seek a certain type of terrain).
One considered solution would be to reduce the number of encounter cards resolved by skipping those that aren’t plausible and taking them as an encounter avoided. While this would reduce the downtime - it would consequently remove the uncertainty and would allow for a meta-avoidance of potential trouble.