Now Allow me to suggest that the user who would attack an Islamic figure would be send a warning. If that happens again then he/she would be banned.
This is, more or less, already the policy across the Stack Exchange network. Moderators will look into any user who has a history of non-constructive (including but not limited to rude/offensive) behaviour. When necessary a user may also be suspended, but this is typically only used if a user has been warned repeatedly, and has still shown no signs of improvement. The user-suspension system is described in detail in Jeff's blog post, "A Day in the Penalty Box."
I feel that implementing a "two strikes, you're out" strategy (as your post implies) would be too harsh; whenever possible, we prefer to discourage problem behaviours rather than discourage problem users. It can be easy to cause offense without intending to, especially when you're dealing with a pluralistic (and public) site populated by users from wildly different countries and cultures. Civility is already required at all times, and rudeness is already not acceptable (regardless of if it's religiously-charged or personal).
Be nice. It's in the FAQ:
Treat others with the same respect you’d want them to treat you. We’re all here to learn together. Be tolerant of others who may not know everything you know. Bring your sense of humor.
What is important to note, however, is that moderators are not all-seeing. While we do attempt to keep an eye on the day-to-day activity of the site, Islam Stack Exchange currently has about 3,500 posts, 2,000 visits per day, and I don't even know how many comments (and that's just for the main site). And (God willing) these numbers are only going to increase every day. That's a lot for four people to keep track of on a regular basis.
The Stack Exchange model relies on a certain level of self-moderation; we the users (and especially we the moderators) can't see everything, but a thousand active users can still see a lot more than four active moderators. One of the first tools any user gets in their toolbox is the ability to flag posts for moderator attention.
Flagging not only lets moderators (and in many cases other high-rep users with access to moderator tools) know that there's potentially someone that we actually need to be concerned about, but it also lets us see exactly what they're doing that's so problematic. We the moderators can't really take action against a problem we don't know about; when there's a history of flagging to refer to, we can actually decide who does (or does not) need warning and/or suspending.
If and when you see something offensive, flag it.