There is nothing at blockchain.info that differentiates between "double-spends" that are an attempt at fraud, and "double-spends" that are simply unconfirmed transactions that get re-sent.
I'm not aware of anyone that has put any effort to carefully determine how many fraudulent double-spend attempts there are. I suppose someone could question all the large businesses that handle many transactions (such as Coinbase and BitPay) and attempt to report some actual statistics, but I haven't seen anything like that published.
Maybe I am wrong, but there is a (small) probability that a possible attacker would "hire" couple of miners to "mine" the double spent transaction before the "real" transaction will broadcast to all the network, having 4+ miners who collect that transaction is less likely.
Confirmations have nothing to do with the number of miners. It is the number of blocks. 1 miner (with enough hash power) can mine 10 blocks in a row, and 10,000 miners could work for years and not mine a single block (if they don't have enough hash power).
If they paid a small amount of money to miner that only has a small amount of hashpower, then the block will almost certainly be solved by someone else. If they want to pay someone that has a LOT of hashpower, then they will need to pay a LOT of money (potentially more money than they are trying to steal). Furthermore, as soon as the first block is solved with 1 of the two double-spend transactions, then victim will immediately know about the fraud. There is no need to wait for additional confirmations. Convincing a miner to throw away their chance at a 12.5 BTC block reward to try and mine a competing block would require the attacker to pay MORE THAN 12.5 BTC for every block that is solved while the attacking miner continues to try. So, if the attacker wants the miner to stop trying to mine legitimate blocks and instead try to replace a block until 4 blocks have been solved, then the attacker is going to need to replace the 50 BTC worth of revenue that the miner is potentially foregoing. All without a guarantee that the miner will even succeed.
This doesn't make any sense. Why would an attacker pay a miner 50 BTC to try and steal 0.1 BTC from a merchant?
From the 100+ transactions I did, I never wait more then 10-15 minutes for 4+ confirmations, only when I add very low fee, or even without fee at all..
(I was wrong, its maybe not 5-6 minutes..)
You weren't paying very good attention to the time.
On average, it will take 40 minutes to get 4 confirmations. If you only did a few transactions, then you might get lucky and get 4 confirmations faster than usual. In that case you might see that your transactions get 4 confirmations in less than 20 minutes. However, you claim to have 100+ transactions, and in that case you either weren't paying attention or are lying.
That having a bigger block will reduce the wait time because there is more "space" for new transactions,
Its also depends on how the new block size will be synchronized with the miners interactions.
Bigger blocks don't result in faster blocks. It will still take an average of 10 minutes for each block to be solved. This means that, on average, transactions will remain unconfirmed for 10 minutes (or more if they pay an insufficient fee), and that once a transaction gets its first confirmation, it will take an average of 10 minutes more for each additional confirmation.