The problem with this kind of situation is that often the subtopic goes through commitment faster than the wider topic. The reason is that a lot of Area 51 participants want to craft a site that caters exactly to their own interests and no more. This is bad, because the resulting small site is less likely to catch the interest of other people. A Stack Exchange site is created not for the sake of the 200-odd committers but for the sake of thousands, potentially millions of users. The right size of a Stack Exchange site is ideally “somewhere around the size of a university department”.
This is a recognized limitation of Area 51, and Stack Exchange is currently thinking how to improve the process. In the meantime, the only way to achive decent-sized sites is to introduce some moderation in the process. Merging proposals is a major aspect of this moderation (and it doesn't get done nearly enough at the moment). Option 3 goes against the implementation of Area 51, not against its purpose.
The problem with overlapping proposals is that committers to one proposal are often unaware of the other, and even when they are, have no incentive to voice their opinion or demonstrate interest. The limited number of commitments also has the unfortunate consequence of making it very costly to commit to two overlapping proposal when you are indifferent as to which one launches.
When a narrow proposal launches, it can leave proponents of the larger proposal in a lurch, because they are now supposed to avoid duplicating the existing site. A site about “X except Y” rarely makes sense.
This isn't to say that it's never a good thing to have a narrow site in addition to a wider site. I'll cite two successful examples: CSTheory is a successful site for researchers in theoretical computer science, and its existence does not preclude that of Computer Science; Ask Ubuntu has more traffic than Unix & Linux which is technically a superset. In both cases, the community of the narrower site could not be served effectively by the larger site. It should always be up to the proponents of a narrower site to demonstrate why a larger site (whether existing or proposed) would not be viable or would not serve their community.
In practice, broad sites are hard to achieve, even with moderator intervention. (Sometimes the problem is a lack of moderator intervention.) So if you find yourself interested by a broad site, I recommend that you go looking for narrower sites, appeal to their users, and appeal to the wider community. In other words, in practice, lobby for the broader site. Remind people of the existence of tags. See who the experts are: would an expert in green Honda Civics know what to do about a red Ford Chevy? If the answer is yes, then the green Honda Civics proposal should be closed in favor of the cars proposal; this is difficult to achieve in the current Area 51 process. (Oh, and the launched site here is Motor Vehicle Maintenance and Repair)