I'm currently developing an HTML5 tablet app for iPad and Android, and so far the biggest limitations are coming not from the HTML5 spec, but WebKit itself. Specifically, there is no way to programmatically trigger focus on form fields, and input fields for some reason do not respond to ontouch events.
Still, I consider it a good gamble long-term. Even though I'll have to degrade gracefully for devices that don't support multi-touch or CSS 3D transforms, it's still less work than trying to maintain two codebases, and I get to use the dev stack I know best.
I tested it on Android, and its WebKit also lacks programmatic focus on input elements. Didn't test the ontouch problem. Mobile Firefox does the right thing.
The funny thing is I intend to package my app in native wrappers and distribute on both App Store and Android Market, so Apple does stand to make money. But, they still lose exclusivity, and it's unfortunate that they have a strong disincentive against improving HTML5 as an app platform.
I dont mean to comment on your situation specifically… I'm simply pointing out you're at a disadvantage:
1. Mobile Safari isn't feature complete
2. Apple has no incentive to make it feature complete
Still, I consider it a good gamble long-term. Even though I'll have to degrade gracefully for devices that don't support multi-touch or CSS 3D transforms, it's still less work than trying to maintain two codebases, and I get to use the dev stack I know best.