The concept is interesting, but I have a couple of problems with it, if I understand the idea correctly. First off, you would have to make it so that it would be impossible to cap the cart in the time that would be added to the timer after capping the cart, or else the map could theoretically never end. * Aside from the obvious problem of being stuck on the same map until an admin forces a change, it also creates much more of an incentive to spawn camp the defending team. For most game modes, the reward for spawn camping is that the main objective becomes easier to achieve. However, for this mode, the reward becomes infinite points, and untold of achievement opportunity, because the spawn camping doesn't end when the timer runs out, because the timer never runs out. Essentially, for all other game modes, the effort spent in pulling off a spawn camp is (Or, at the very least, should be) actually greater than that needed to win the round, and thus not really worth the effort, since the reward is the same. However, for this mode, it would still take more effort to pull off the spawn camp, but the reward would also be infinitely greater, thus making it more than worth it. That "infinite" back there isn't just for drama, either, it really would be true. Red would be able to cap the cart forever, meaning they would be getting points forever, until an admin either breaks the spawncamp, forces a map change, or until the massive rage quit domino stack empties the server**.
The next problem is the circular nature of the cart path. Basically, no matter what the shape of the cart ends up being, the cart path will still have a "Farthest point from attacker's base" and a "Closest point to the attacker's base", effectively creating a circular or oval shape. Now, on a standard payload map, Red is always trying to push the cart away from their base and into (Or at least towards) Blu's base. And this is exactly how this map type would play out during the part where you get from your base to the defender's base. However, a problem occurs when you reach the part where you get it from their base back to your base, so you can start the next lap around the track. When Red pushes the cart AWAY from their base, it becomes harder to continue pushing it***, because every foot you push it is a foot farther from your spawn and resupply, and a foot closer to the enemy's spawn and resupply. However, when you're bringing the cart back to your base, the exact opposite happens. The cart is getting closer to your base, so it becomes easier to cap the cart as time goes on. This is exceedingly unfair to the Blu team, as it takes away practically all advantages they have as the defending team, leaving them powerless until the cart starts being pushed back into their base. It's essentially as if, on Fastlane, after you cap the middle point, the objective then becomes to cap your own second point, and then your first point. It doesn't get harder as time goes on, but instead starts out hard and then gets easier.
Another good question to ask is "What does this add something to the game that regular payload doesn't?" Let's take your Badwater Basin example. You take the cart over to the first cap point, just like any regular game of Badwater, and then you bring it back to your spawn over the hill. (Or via another route, it doesn't really matter. The point to this will still be the same) Now that you've brought it back to Red's base, what happens now? The answer is, the exact same thing. Rather than a new type of payload, it's instead regular payload played on a smaller map over and over again, with little prospect of teams being switched. And I have been speaking about the one-cart mode this whole time, but this also applies to the two-cart mode. It's not really different than plr_, it's just the same plr_ track being played over and over again.
In short, the one-track mode is too conducive to spawncamping, too repetitive, and all around similar to regular payload, but with even more bias against the defending team than regular A/D gametypes. The two-track mode is interseting, as I like the idea of both carts being on the same track, rather than two opposing tracks, but again, it is very similar to regular payload race. Nerdboy's point about dividing the teams is also valid, especially for the two-cart mode, since it essentially taes away the incentive to stop their cart when you can just pile up pain trains and scouts on your cart, since the circular nature would keep the carts from coming into conflict. (Basically, the cart would always be about equidistant from each other, rather than passing each other like in Pipeline)
I will say, though, that there are two things I like about this concept. The first is the basic idea at the center of this. It basically abandons the idea of "Push leetle cart to this point, so you can get more time to push it to the next point, until you get it to the last point" and instead adopts the mindset of "Here's a large chunk of time. There is no end destination, just see how far you can get your cart in the time limit." It's like if, on Cashworks, instead of getting ten minutes at the start, and five more every time you cap until you get the last cap, you instead started with thirty minutes, and the goal was to see how close you could get to the giant money vault at the end. If you make it to the vault, awesome, but if not, that's okay. The idea wasn't to get to the end, but to see how far you could get from the start.
The second thing I like is that it would allow us to see how a team changes it strategies as the round goes on, without changing the geography of the area every time a new round starts. Essentially, with Goldrush as an example, Blu can fail to defend the first point. After they fail, they can think about why they failed, and what they could have changed so that they kept the first point, but it won't matter, because now they have to defend the second point, which has a whole new set of geographic features to utilize, and by the time they defend that first point again, a couple rounds later, they've forgotten the lessons from their failure, and have begun to employ a different version of that same failing defense. However, with this game type you've made, They could still fail, and lose that first point, but it's alright. They'll be defending that same point again in just a few minutes. Any lessons they learned from their recent failure can be implemented, and if they fail, they can just keep trying again until something works. And once the attackers are stopped, they have to use the surplus time they've gained from running the cart around to try new attacking strategies to break the defensive line, until they succeed, or the timer runs out. I can't say it would be fun to play in that kind of round, but it would be interesting to spectate it.
* Assuming, of course, that the server is set to change maps at the end of the round, rather than when the map change timer runs out.
**Essentially, Blu will start to ragequit from being camped. Then, Red team members will get autoswapped to Blu. The newly autoswapped members will gain a newfound appreciation for why spawncamping sucks, and then will ragequit. The process repeats itself until the server is empty.
*** I realize that once you destroy the defending teams initial setup, it becomes easier to push the cart forward, even though it's farther from your base than it was before. However, for the purposes of argument, I am assuming an equal amount of resistance during the whole length of the round.