So I am not sure what you are getting at here.
In the current model, you have altcoin miners, which are selling their altcoins, just to get some BTCs. Because it is more profitable to mine some altcoins, and sell them, than to mine BTC directly. Which means, that first, you get some altcoins (weak blocks on layer two), and then you convert them to BTCs (stronger blocks on layer one). In between, your altcoin can lose or gain value, it can be unstable, there could be some timing attacks, some speculation attacks, and a lot of other attacks.
So, why not have a proper layer two, instead of having weak altcoins, and selling them, just to get some BTCs? As you can see, those who sold their LTCs, XMRs, or other coins in the past, and converted them into BTCs, they gained more, than those, who simply kept their altcoins. And of course, there are cycles, there are times, when people buy a lot of altcoins, and there are times, when they sell a lot of them. But as long as BTC domination is jumping around 50%, it is the same situation, as having 100% domination, and 42 million coins, but locked in different networks.
No serious entity would use Bitcoin this way.
There are many parties, which accept a single BTC confirmation as "rock solid". Also, many sites will show you, that some transaction is "pending", if it will have zero confirmations, and will be recorded in their mempool. As long as your amount is "coinbase covered", you are good to go. Which means, if the coinbase amount is for example 3 BTC, and you are sending 1 BTC, then you won't reorg this "3 BTC block", and replace it with something else, because it will cost you more than 1 BTC, to successfully execute a double-spending attack.
Obviously, "coinbase coverage" also works in altcoins. Which means, that if the altcoin reward is 50 ALTs, and you are sending 1000 ALTs, then you should wait something around 20 blocks (1000/50=20). To get exact numbers, you can of course compute Poisson distribution, but using "coinbase coverage" is just easier: for amounts smaller than the coinbase, a single confirmation is usually sufficient, and for amounts bigger than the coinbase, this rule will usually give you bigger amount of confirmations, than you need.
Again, I really don't understand what you are talking about here.
Using more than one coin is not that different, than using more than one layer. If you have some altcoin, you mine it, you sell it, and then you get some BTCs out of that, then it is not that much different, than having a proper second layer, instead of using two coins on the first layer.
The main difference, is if you have layers, then you can have a strong peg, for example 1:1 peg. However, if all coins are on their first layer, then there are different exchange rates, there are different network conditions, and you risk more, when converting between BTCs and ALTs, than if they are all connected, through some kind of layers.
For example: if you use LN, then you lock your BTCs, you get 1:1000 constant exchange rate, and everything is converted from satoshis to millisatoshis, and then you can convert it back, using 1000:1 constant exchange rate, to get all of those locked satoshis back. However, if you do the same with BTC and ALT, then it is no longer 1:1, 1:1000, or some other constant rate. It is volatile. And those networks are different, there are different risks, different network rules, timing attacks, and so on. And by rejecting maximalism, and second layers, every coin uses just their first layer, and it is much harder to make a risk-free trade (because even if your ALT fees are smaller, then the price of your ALT can change in the meantime, and it may turn out, that BTC fee was higher, but the price change was smaller, so the final outcome was better).