[TextEdit either sl nore ml][fr_BE] Scancode (0x1a 0x9a) issue

```
[dependencies]
rustc_version_runtime = "0.3.0"
text-colorizer = "1.0.0"
clear_screen = "0.1.0"
eframe = "0.33.3"
egui = "0.33.3"
lazy_static = "1.5.0"
regex = "1.12.3"
```
I use the TextEdit widget like this...
```
let commentaires=elements.add(TextEdit::multiline(...).desired_rows(12)
.background_color(...))
.text_color(...)
.char_limit(....into()));
if commentaires.gained_focus()
{
self.strComment = "".to_string();
}
```
When I use my application there is a big misfunction using the '^' feature on french keyboards to have the circomflex accents on 'i' on 'u' or 'e' for example...
Either multiline or singleline you could only put the circomflex on the first character when using scancode(0x1a 0x9a)... each other '^' would NEVER be printed...

Type a text either in single line or multiline and try to put a '^' (scancode 0x1a 0x9a) on layout fr_BE on the second circomflexed character in this sentence...
-> ô combien de temps ? Ces êtres ne sont pas fiables... ces îles sont maudites !!
The second 'ê' and the third 'î' won't display on the widget...

Operating System: openSUSE Leap 15.6
KDE Plasma Version: 5.27.11
KDE Frameworks Version: 5.115.0
Qt Version: 5.15.12
Kernel Version: 6.4.0-150600.23.87-default (64-bit)
Graphics Platform: X11
Processors: 16 × AMD Ryzen 7 3700X 8-Core Processor
Memory: 62.7 GiB of RAM
Graphics Processor: AMD Radeon RX 550 Series
Product Name: B550 Phantom Gaming 4

The entry point is egui's TextEdit widget, using the supplied singleline and multiline examples with a fr_BE keyboard layout; first reproduce the missing second circumflexed character. Trace the widget's text-input handling and verify repeated circumflex composition; done means the second ê and third î render correctly in both modes.

Written by the indexing model from the issue text.