Quick answers-links to the most frequently asked questions.
Since Gearbox Protocol has two sides to it, the fees you are paying depend on what user type you are. If you are a passive Liquidity Provider, there is no direct fee for becoming one.
Diesel Tokens are like cTokens of Compound. These tokens automatically earn interest & fees proportional to your share of the pool like cTokens on Compound or Yearn LP tokens. You don’t need to claim interest or perform any other actions, your Diesel Tokens value is supposed to grow.
Depending on the utilization ratio of pools, if it is high - not all the liquidity can be withdrawn at once. It doesn't mean that the protocol lost money or is insolvent - it's just that the liquidity is utilized in Credit Accounts. In such cases where available liquidity is low (which you can check in the app analytics page per pool) - the APY goes higher as to attract more capital providers and attempt to find an equilibrium. This is more or less standard among all lending-type protocols and has happened both in Compound & Aave before. Gearbox DAO can adjust the utilization curve parameters to force certain behavior of the protocol in case weird situations occur.
As the protocol has max leverage and max personal borrow parameters, which are configurable by DAO, there is a limited amount you can borrow for leverage. As such, borrowing above that amount is technically not possible. Once a Credit Account is open, you can add more collateral to increase your position size and Health Factor, but you can't borrow to infinity for security purposes.
As the protocol has gone live just recently - nobody knows what PMF (product-market fit) Gearbox has. The app will likely take many iterations, and anybody can build their own. So while this search for users and market fit are ongoing, the effort on optimiizing 100 versions at once is minimized. Later on, as things become more clear - the mobile version will be optimized and turned on.
Gearbox Protocol is DAO-operated, so it should be a joint community decision. That's one. If users need Gearbox, it's fine to deploy anywhere, because user-first approach.
But it's not practical being on chains where there are not many other protocols to tap into. You see, Gearbox sources liquidity from other DeFi protocols: trading on Uniswap and Curve, farming in Yearn & etc. If a chain only has a Uniswap (original or fork) and a Curve (original or fork) it is  less safe usually  what is more important - it makes the product of Gearbox be inferior. If there is not much liquidity or you can't build composable positions, there isn't much to do on that chain.
So in the future there is no worry with it, but current ecosystems need to develop more and have more safe and liquid protocols to justify time and development costs. Which is likely possible, and is something to altogether discuss in 2022. See some ideas at the bottom of this article.
Because ETH is not an ERC20 token by itself, so you can't operate it like an ERC20 token which is a must-have for Gearbox Protocol and safe operations. Within the Credit Account, ETH is converted to WETH. So if you connect via Wallet Connect or see your Credit Account on Etherscan - and check for ETH, you will see WETH only. After liquidation or closing your account, it is converted back to ETH, so liquidity providers don't need to think through this too much.
Due to how Threshold Weighted Values are calculated taking into account a possible price drop and a price feed tick, a stablecoin is not actually a stablecoin for the protocol contracts. They are not 1:1. Stablecoins tend to de-peg, which must be taken into account. This keeps the protocol more secure, but also causes max leverage parameter to be lower. As such, this is up to the risk appetite of Gearbox DAO to decide on such things.