mruby-regexp: drop the private UTF-8 encoder in favor of mrb_utf8_to_buf - #7094

The regexp gem carried a private UTF-8 encoder, mrb_re_utf8_encode in
re_utf8.c, whose implementation is byte for byte the same as
mrb_utf8_to_buf in src/string.c. This PR deletes the private copy and
switches its only caller, emit_codepoint in re_compile.c, to the
shared function.

Using the shared encoder from a gem follows existing practice:
mruby-sprintf, mruby-pack, and mruby-string-ext all call
mrb_utf8_to_buf through <mruby/internal.h>, and regexp.c in this
gem already includes that header.

The one difference between the two functions is that mrb_utf8_to_buf
returns 0 for a codepoint above U+10FFFF, where the private copy
assumed the caller had validated its input. The compiler satisfies that
assumption: every codepoint reaching emit_codepoint comes through
check_unicode_cp, which rejects surrogates and anything past the last
plane as "invalid Unicode range" before encoding, matching CRuby:

/\u{110000}/   # regexp.rb:1: invalid Unicode range (SyntaxError)

So the zero-length case is unreachable and no caller changes behavior.

Net effect: 3 insertions, 31 deletions, no functional change.
rake test passes (2050 tests, 0 failures; bintest 105 OK).

Summary by CodeRabbit

• Refactor
  • Streamlined internal Unicode handling in regular expression processing.
  • Updated Unicode encoding references to use the runtime’s standard implementation.
  • Removed redundant internal encoding logic without changing user-facing behavior.