Fix ascii_only? flag in strio_write

[tompng]

Fixes https://bugs.ruby-lang.org/issues/20185
Followup of #79

Looks like rb_str_resize is changed in some ruby version

rb_str_resize(string, shorter) // clear ENC_CODERANGE in some case
rb_str_resize(string, longer) // does not clear ENC_CODERANGE anymore

// rb_str_resize in string.c
if (slen > len && ENC_CODERANGE(str) != ENC_CODERANGE_7BIT) {
  ENC_CODERANGE_CLEAR(str);
}

I think this change is based on an assumption that appending null bytes will not change flag ascii_only?.

strio_extend will make the string longer if needed, and update the flags correctly for appending null bytes.
Before memmove, we need to rb_str_modify because updated flags are not updated for memmove.

[kou]

Wow.
ruby/ruby@b0b9f72 may be related.

[kou]

Suggested change

	refute(s.string.ascii_only?)
	assert_operator(s.string, :ascii_only?)

[tompng]

changed to refute_predicate(s.string, :ascii_only?)

[kou]

Oh, sorry.
Could you use assert_not_predicate instead? I don't want to use refute_* as much as possible because assert_not_ is straightforward for non-native English speaker.

[kou]

Oh, @nobu worked on this: #79

[tompng]

I add a test case

    s = StringIO.new("\u{3042}")
    s.rewind
    assert_not_predicate(s.string, :ascii_only?)
    s.write('aaaa')
    assert_predicate(s.string, :ascii_only?)

that fails in master (after #79)

Writing string in arbitary position(when ptr->pos != olen) can change coderange in many variation, and I think rb_str_modify is always needed.

Other variation:
writing "\x00".force_encoding('utf-16be') twice can make string invalid to valid
s.write "あ"; s.rewind; s.write('a'); s.string #=> "a\x81\x82" will break encoding.

[kou]

Thanks!