Continued lists are majorly broken after transformation

[Vampire]

If you transform this:

- foo
    - bar
        - baz

      bam
      boo
    - bae

you end up with

* foo
** bar
*** baz bam
 boo
- bae

which is semantically totally different.
"bam boo" should continue the "bar" point after the nested list containing "baz" and "bae" is a sibling to "bar".
With the current output "bam boo" is joined to "baz", producing "baz bam boo", and "bae" is nested inside "baz bam boo".

(When you try to get it right, don't get fooled by Asciidoctor also parsing lists incorrectly,
with the correct AsciiDoc result Asciidoctor renders "bam boo" as continuation of "foo" and "bae" starts a new list [relevant with ordered lists], but that is an Asciidoctor bug that you can easily work-around though manually)

The correct result that also kramdown-asciidoc produces (besides that it indents the list markers) would be:

* foo
** bar
*** baz

+
bam
boo
** bae

[bodiam]

I investigated this issue and unfortunately it's a fundamental limitation of the pegdown parser used by this project.

When parsing the example input, pegdown produces an incorrect AST — it merges the continuation text ("bam boo") and the subsequent list item ("- bae") into a single text block inside the parent list item, rather than correctly representing them as separate structures. Since the AST is already wrong at parse time, there's no way to fix this in the serializer.

Fixing this would require replacing pegdown with a more capable Markdown parser that correctly handles list continuations after blank lines in nested contexts.

For reference, I was able to fix 6 other open issues (#60, #62, #63, #64, #127, #128) — see the corresponding PRs.