Re: [PATCH/RFC 0/3] Support ignore rule "**" using wildmatch from rsync

I skimmed the series and I think it would be a good change in the longer
term, but I suspect we probably have an unfinished business on the core
side that sometimes uses pathspec->raw field and assume that they only
need to do the leading-path match, which needs to be fixed first before
changes like this can be sanely added.