The 10 minute blocktime of BTC prevents its usage directly for many use cases. So if a 1:1 price peg to BTC is required and the 24+ hours latency is not a problem, then it seems a side chain can have whatever characteristics desired. It is definitely an interested method.
Do you see a time when this blocktime could be reduced?
The 10 minute block time does not prevent its usage.
A shorter block time means an easier puzzle to solve which in turn means more confirmations are needed to have certainty that a double spent won't happen.
Large values, you wait.
Small values, as long as the TX fee isn't abnormally small, seeing the TX is enough to have high confidence it won't be a double spend. A block reward is worth $10k and will be worth $5k after the halving (at current rates), the only double spend attack that don't risk losing a block reward is a race attack where two transactions with the same inputs are put out on the network at the same time and sending the one you hope loses the race to the blockchain directly to the node the vendor uses.
That can be defended against by not accepting connections from other nodes and making sure your node connects to well connected nodes. Doesn't make race impossible but makes it extremely difficult.
BitPay accepts transactions with 0 confirmations and instantly the vendor indicates it is paid and I get my goods. The time between blocks isn't an issue since most attacks that aren't race require risking a very valuable block reward.
Reduce the time and not only do you make it easier to mine a block and do these attacks, but you reduce the financial loss if the attack fails because the block reward would also be reduced.
Bitcoin's time between blocks does not need to be changed and should not be changed.
Making micropayments easier on sidechain so it doesn't use space in the block however is a change worth considering.