how to set this voice-crossing situation ?

[Keith OHara]

Dear LilyPond users,
Ryan posed a question on the bug-report list regarding how to set the example 
in the attached image, with the repeats written out.

You see the new voice is written conceptually as the upper voice, although it 
is lower in pitch than the voice being repeated in the percents.  We have a 
voice-crossing situation.

Straight-forward input in LilyPond crosses the stems of these crossed voices, 
but we can override to avoid that.  Any better suggestions ?
        
\transpose c c'' { \time 2/4
  e8[ a e a]  e8[ a e a]
  <<
   { % \once \override  Beam #'positions  = #'(-1.0 . 0.8)
     e8[ a e a] }
   \new Voice {\voiceOne r4 r8 fis,8 } >>  }

Sample.png
Description: PNG image