a user that download and using a script should have so much knowledge so they know the version of mirc is 6.31 and not 6.17
This is also an idealistic view. The truth is, though, that most users who are new to any community or program are rarely aware of the latest releases and will therefore trust the most recommended source. If a very popular script recommended by their good friend is being packaged with 6.17, the user probably won't question it. The user shouldn't be faulted for something like that.
the author of a script care for his users, then he should let them know that the script using a outdated version of mirc.
Therein lies the problem. Script authors rarely put much attention into the script packaging. They'll toss everything into a zip and you'll be lucky if you even get installation instructions, though review-based scripting sites have done a good job in changing that mentality.
But reality aside, even if authors cared about their user base, there would be nothing holding them to actually acting on it and informing their users that the installer may be old. Some authors might
care, but may inadvertedly omit the fact that the mIRC packaged with their script could very well be over a year old. This can easily cause problems if the script author is, say, the author of PnP or other. This is why the license agreement states clear rules about how script authors can and can't distribute mIRC.
Frankly, it's simpler as a script author to deal with the packaging of their own script and allow the packaging of mIRC to be handled by mIRC itself. I don't even see much of a benefit to bundling. It would be very easy to just inform a user to visit http://www.mirc.com/get.html
and download the program on their own. You could even launch the browser for them, as Riamus suggested, and there is no violation of the EULA in even going as far as automatically downloading the installer from mirc.com using a custom script installer of your own. It's not like bundling will bypass the installation process of mIRC-- it would only bypass the downloading part-- is clicking a link *really* that difficult for a user? I personally don't think one less browser click would justify all the extra complications.