runtime(syntax-tests): Break up non-ASCII over-75-byte-long lines

And anticipate occasional multibyte line wrapping owing to:
> A poorly rendered line may otherwise become wrapped when enough of
> spurious U+FFFD (0xEF 0xBF 0xBD) characters claim more columns than
> are available (75) and then invalidate line correspondence under test.

Observe that for "vim_ex_command.vim" another workaround is
chosen: the long line containing an only multibyte character
near its EOL is conversely made longer by padding and moving
the character to a separate _tail_ part of the wrapped line.
That is, the _head_ part of the line is all ASCII characters
and the wrapped _tail_ part is a mix of various characters
whose total byte count is within bounds.

Other unmodified tracked files of interest:
  java_lambda_expressions.java,
  java_lambda_expressions_signature.java,
  java_numbers.java,
  markdown_conceal.markdown,
  vim9_generic_function_example_set.vim

Also, remove stray U+FFFC (0xEF 0xBF 0xBC) characters.

Related to #16559 and #17704.

Reference:
0fde6aebdd/runtime/syntax/testdir/README.txt (L120-L123)

closes: #17868

Signed-off-by: Aliaksei Budavei <0x000c70@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>

runtime/syntax/testdir/input/java_methods_indent2_signature.java

@@ -67,8 +67,8 @@ abstract class Indent2$MethodsTests
   protected static final synchronized strictfp void ascii$99_()
   { ascii$98_(); }
 
-  @SuppressWarnings("strictfp")
-  protected static final synchronized strictfp <α, β> Τʬ<α>[] μʭʭ$99_(
+  @SuppressWarnings("strictfp") protected
+  static final synchronized strictfp <α, β> Τʬ<α>[] μʭʭ$99_(
                         java.util.function.Function<β, Τʬ<α>[][]> ƒ)
   {
     return