The pegging will be done on-the-fly in-and-out as needed by the users. It will only need a small amount of bitcoins to start. If a better SC comes out, you can switch back to your bitcoins and switch to another SC anytime.
If someone builds the uber SC, then a lot of people will switch, but the bitcoins will still exist just in case a better SC comes along someday. Bitcoin will not be destroyed because it is the primary liquidity for the SC. In fact it should drive the price of BTC up for cold storage purposes just in case the SC gets broken. Miners will be the sentinels for Bitcoin.
What if miners migrate to the sidechain diverting mining resources from bitcoin. Would this leave bitcoin exposed to a 51% attack? If a 51% attack were easy, then some of the conversion transactions could be reversed, leaving an imbalance of coins and screwing up the peg.
Edit: I guess mining would be more or less pegged too. Miners could mine bitcoin, convert so SC and sell, at least as long as there was a mining reward.
if i was a gvt, i'd fork the mainchain into an uberSC that employed a single beneficial factor over the mainchain, anonymity being the first one. now, as BldSwtTrs has said above, all current mainchain BTC tucked away in cold wallets all over the world will probably be cracked open out of greed and some out of necessity and sent to the uberSC. even tho the security at the moment might be risky, why not, as the paper has said that even in a SC failure you can get back into your BTC via the peg. all upside, no downside! which is it Greg, can we get back or can't we?
assuming the paper is accurate, mining will be forced to transition/follow as well, but that is where the security holes start gaping wide open. as you said, attacks will made on the network causing double spends and loss of confidence. also, individual hodlers making the transition now risk identification and hacks.
now, to amplify the attack as the gvt, i would deploy as many new single feature uberSC's as there are beneficial features with a divide and destroy objective. BTC and miners would be scrambling to different uberSC's they perceived gave them the most benefit only to be eventually destroyed by attacks as a result of less security for each uberSC during the transition.
this is why, potentially, all innovation should be done on the mainchain with the current security model, hashing rate, and BTC distribution in place safely tucked away.