Ancient accidental inconsistencies?; Morgenlied quesitons.

[Edward Sanford Sutton, III]

        http://lilypond.org/doc/v2.1/input/test/out-www/lily-971630970.ly 
(ancient-accidentals.ly from http://lilypond.org/doc/v2.1/input/test/out-www/
collated-files.html) on the website appears to show the same doubleflat 
accidental across all styles (which appears to be the same as the default 
doubleflat). Is this correct?
        Are there any accidental choices to make a doublesharp with the dots to 
its 
left and right, or one with dots on all four sides?
        Is this the proper place to mention bugs for cvs lilypond versions, or 
is 
bug-lilypond preferred for that too?
        In http://lilypond.org/doc/v2.1/input/mutopia/F.Schubert/out-www/
morgenlied.pdf on the end of the first line in the lyrical staff, the 
extender line coming off of du appears to be smaller than a period. Is there 
currently, or should there be a limit to valid extender line lengths?
        In the same piece, within the vocal line, there are a lot of stem down 
notes 
despite when near the bottom of the staff. Was this a slip caused by double 
voicing in measure 3, an attempt to clone the original, or desired for other 
reasons? If still desired (or cloning), why should they go down? I'm still 
attempting to get my hands on another score to compare to myself.
Thanks again,
Edward Sanford Sutton, III