This fixes two bugs in the parseJSDocCommentWorker().
1. The initial indent was calculated wrongly. It was set to the
difference between the index of the last newline or beginning of file
and the current start marker (position of /**). By calculating it
this way, the newline character itself is counted as indentation
character as well. The initial indent is used as margin for the
whole comment. The margin contains the amount of characters to skip
before the actual content or payload of a comment line. The algorithm
does not skip non-whitespace characters at the beginning of the
content, but it would strip away one whitespace character for
indented content (which does matter, if there is e.g. a Markdown
code block with indentation in the comment).
2. When reducing initial whitespace sequences of comment lines by the
remaining margin the algorithm cut off one character too much. This
might have been introduced to fix 1. It had a similar effect as 1.