regex does not properly select match

[Benjamin Eeckhout]

The text selection when searching in a html file does not seem to be correct.

This is a mininal reproducable example:

```
𝒎𝒎𝒎𝒎𝒎𝒎
<p> test</p>
```
with the regex:
```
test(?=<\/p>)
```
I verified it on `https://regex101.com/` that this generates different matches (it selects `test` as expected)

I really don't know _what_ it is doing exactly, depending on where the search starts + it gives different selections if `up` or `down` is set as search direction.
Sometimes it _does_ give the correct match which makes it even more confusing.
It does **not** exhibit this behaviour when using "normal" characters (eg: "a", "b", ...) and breaks with "weird" characters (eg: "𝒖", "𝒎", ...).

Version: 7.16
OS:
Edition Windows 10 Home
Version 22H2
Installed on ‎28/‎12/‎2023
OS build 19045.4651
Experience Windows Feature Experience Pack 1000.19060.1000.0

[Benjamin Eeckhout (benjamineeckh)]

It looks like this happens if the regex parser goes over any non-single byte character while traversing the file.

So with search direction as "down" and `|` being the cursor position:
```
𝒎𝒎𝒎𝒎𝒎𝒎 |
<p> test</p>
```
with the same search will correctly match `test`, while:
```
|𝒎𝒎
<p> test</p>
```
will match "> te" (probably because of the 2 extra bytes).

Also note:
search was in "regex" mode not case sensitive , wrap was on and "dot all" was off

[Kovid Goyal (kovid)]

Fixed in branch master. The fix will be in the next release. calibre is usually released every alternate Friday.