Regex positive lookahead - pattern that always fails

Question 1

No, there's no matching pattern similar to the one you mentioned that I have seen, but such a structure can be used (though admittedly a bit weird), for example:

q(?=.*t)u

This regex will match any string starting with qu but which has a t somewhere later. This means that question, quit will match, but not quasar. In which case, an equivalent and more readable (imo) regex qu(?=.*t) could be used.

Question 2

The regex lookaround syntax is zero width.

What this means is that it matches but doesn't move the cursor, so in your pattern:

q matches "q" the cursor moves to "u"
(?=u) matches "u" the cursor stays at "u"
i does not match "u" so the pattern fails.

Note that the pattern doesn't backtrack, the lookaround assertion is zero width.

The structure is very useful is you wanted to match a pattern that contained, say, "at least one X" in a series of letters. For example:

[a-z]{4}[1-9]{3}(?=.*X)[a-zA-Z]{5}

Says four lowercase letter followed by three digits followed by five letters of any case with at least one "X".

Question 3

I would say, if the pattern after(or before) the lookahead ((?=...)) is fixed. the regex doesn't really make much sense. like:

foo(?=bar)fixed

but if the fixed part is dynamic, it would be useful. see this example:

kent$  echo "fooququuuxxxxxxx"|grep -Po 'q(?=uu).*' 
quuuxxxxxxx

kent$  echo "fooququuuxxxxxxx"|grep -Po 'q(?=u).*' 
ququuuxxxxxxx

in above example, only the lookahead is different, you got different match result.

Question 4

In applying the use case structure presented in this question and the response by Boris the Spider, I came up with this solution

$detail = '[code]<!doctype>
<html>
  <head></head>
  <body><p>My Regex script</p></body>
</html>[/code]';

function regex($detail) 
{
   if(preg_match('#^\[code](?=.*(<([A-Z][A-Z0-9]*)\b[^>]*>(.*?)</\2>))\[/code]#si',  $detail))
   {
       return true;
   }
   return false;        
}
echo regex($detail);

Taking a look inside the regex engine, this is what is happening. In applying ^\[q](?=.*(<([A-Z][A-Z0-9]*)\b[^>]*>(.*?)</\2>))[/q]* to $detail above; \[q] matches [q] and the html section of the code matches (<([A-Z][A-Z0-9]*)\b[^>]*>(.*?)</\2>)).

The match from the lookahead is discarded, so the engine steps back from [/q] in the string to the html section of the code. The lookahead was successful, so the engine continues with [/q]. But [/q] cannot match (<([A-Z][A-Z0-9]*)\b[^>]*>(.*?)</\2>)). So this match attempt fails.

However this regex synthax works:

#^\[code](?=.*(<([A-Z][A-Z0-9]*)\b[^>]*>(.*?)</\2>\[/code]))#si

The regex engine simply says: "opening [code] tag followed by any character, then a pair of html tags with at least a closing [/code] tag at the end".

I hope this helps explain the use case pattern match (quit) more.