To allow clients to influence the number of comment prefix matches; especially when passing a line range for comment detection, the comments may not appear uniformly or at all.
Now actually just return the pattern for the indent (+ comment prefix), without capture groups and a full match for the remaining text. This is what clients actually need.
Implementation in s:SplitIndentAndText() was based on unnestable prefixes. Instead of adding 1 to the actual nesting level to obtain the pattern multiplier, ensure it is at least 1.
Change its implementation to just erase the prefix itself.
Add ingo#comments#SplitIndentAndText() to provide what ingo#comments#RemoveCommentPrefix() was previously used to: The line broken into indent (before, after, and with the comment prefix), and the remaining text.
FIX: Wrong negation in a:options.isIgnoreIndent documentation.
CHG: Make a:isIgnoreIndent flag to ingo#comments#CheckComment() optional and add a:isStripNonEssentialWhiteSpaceFromCommentString, which is also on by default for DWIM
CHG: Don't strip whitespace in ingo#comments#RemoveCommentPrefix(); with the changed ingo#comments#CheckComment() default behavior, this isn't necessary, and is unexpected.
Ingo Karkat