fix: grow NgxString on write! instead of failing within capacity

[u5surf]

Proposed changes

Closes #326, taking the fix you preferred there.

 fn write_str(&mut self, s: &str) -> fmt::Result {
-    self.append_within_capacity(s).map_err(|_| fmt::Error)
+    self.try_append(s).map_err(|_| fmt::Error)
 }

try_append reserves before it appends, so allocation stays checked: a genuine
failure surfaces as fmt::Error rather than a panic, which is the property
String does not give. It also means a failed write leaves the string
untouched instead of holding the part that fit.

No separate fixed-size formatter. append_within_capacity is unchanged and
still public for anyone who wants that behaviour directly.

Testing

Two tests added:

• write! into a string with no capacity reserved now produces the whole
  value, which previously failed on the first byte.
• write! backed by an allocator that always refuses returns an error, does
  not panic, and leaves the string empty.

The existing test_string_write is untouched and still passes. It reserves up
front and asserts the buffer pointer and capacity do not change, which holds
because try_reserve_exact does nothing when the spare capacity already covers
the write. So the reserve-once-then-fill pattern in examples/shared_dict.rs
keeps its single allocation.

Checklist

• I have written my commit messages in the Conventional Commits format.
• I have read the CONTRIBUTING doc
• I have added tests (when possible) that prove my fix is effective or that my feature works
• I have checked that all unit tests pass after adding my changes
• I have updated necessary documentation
• I have rebased my branch onto main
• I will ensure my PR is targeting the main branch and pulling from my branch from my own fork

[bavshin-f5]

Looks good, thanks!