Non-alphabetical start of the string breaks the markup in some cases

Good morning,

I'm the author of the Jira ADF -> Markdown converter(not published yet). And I faced the next issue in some advanced cases. The easiest way to describe such cases is to provide the examples:

**bold*****bold italic***   // works correct
**bold** ***|bold italic*** // works correct
**bold*****|bold italic*** // works incorrect
**bold****​*​bold italic​*​** // works incorrect

The problem is when there are several similar rules one-by-one without any breaks, the non-alphabetical symbol at the start of the string breaks the markup, even though it's not used for any rules.
In the last example I used a Zero-width space to be sure that there are no rules used for such symbols.

I would appreciate it if you resolve this bug.

The issue names no files or tests. Reproduce the four markup examples in the JavaScript parser, then trace inline-markup handling; done means the failing adjacent-rule and zero-width-space cases parse like the working examples, with regression coverage for them.