How does creating other currencies abuse the 21 million coins limit?
If you have only one cryptocurrency, with a limited supply, then the limit is enforced properly. If you have two cryptocurrencies, and each of them has 21 million coins limit, and also each of them has 50% of the market, then there are 42 million coins on the market. If both chains have 1:1 peg, and you can do exactly the same things on each coin, then you have 42 million coins, and you don't have to worry, which one you are currently holding. Then, you are in a situation, where the end result is exactly the same, as if you have a single network, but with 42 million coins.
And then, it doesn't matter, that you can say "I accept only coins from network A, but not from network B". As long as 1:1 peg is maintained properly, users will find a way to swap, and give you exactly the coins you want. Also because there will be two camps, and someone else will accept "only coins from network B, but not from network A". And if you will try to actively enforce the limit of your favourite network, then people will call it "censorship".
What if I want my network to not be dependent on transaction fees but on tail emission?
You can do that, but it doesn't change the fact, that by doing it, you are creating additional inflation. Instead of tail supply, you could also grab single satoshis from each UTXO, and enforce it by your consensus rules. The whole economy of your network would be identical, but tail supply is just a more obscure way of doing that. If you have a limited supply, you can actually see, how you gradually lose your coins. If you have no limits, then numbers stay the same, but they are worth less, so you can purchase less products and services, for the same amount.
Or, what if I don't want my network to be dependent on another network? (In this case, Bitcoin.)
Then, other people will make swaps, and it will be dependent. For example: testnet3 was supposed to be worth zero, and independent of the main network. Now, it is actively traded, and even though some people are trying to kill it, the outcome will depend mainly on users: if they will want to keep it alive, then it will live as an altcoin.
If you have two properly spinned networks, then it is cryptographically possible to always trade between them, without trusting anyone. And of course, a lot of centralized services will be built as well, if users will want that.
What if I want my network to be ASIC resistant?
Then you should be at least honest in your consensus rules, and set "a difficulty limit", where after reaching it, produced coins are invalid, and for example the whole chain stops (or: rather a chainwork limit, that would avoid some difficulty attacks, where powerful miners can produce weak blocks, and keep the network running, despite breaking the rules).
Surprisingly, "ASIC resistance" is probably not what users really want. For example: testnet3 and testnet4 are networks, where CPUs and ASICs can co-exist. And this is probably more close to what people tried to achieve. Because then, ASICs still can secure your network, but you, as a developer, can still mine your blocks on your CPU. And then, some ASIC can burn a lot of energy, to produce 50 coins, while you can spend a fraction of that energy, to produce the same 50 coins on your home computer. Then, the end result is a network, which is not "ASIC-resistant", but where ASICs are basically discouraged, because "why produce a block with 64 leading zero bits, where you can get the same reward for 32 leading zero bits".
Does that really make sense? If I sit in my warm sofa, if you sit and we only buy Bitcoins, how the hell will our life improve? We have to get up and start working to improve our life, Bitcoin is an empty coin if everyone lays in a warm sofa.
.
For real, though, that was a nice addition. 
