Linearity Check in fixed ecalgeo.g

Abstract:  A bug in the EEMC geometry has been identified.  Lines such as "Material Air isvol=0" have an unexpected (read undocumented) behavior: they create a new material ECAL_AIR with the properties of the last defined volume.  In the EEMC geometry, this results in ECAL_AIR taking on the properties of Iron.  In this blog entry, I test a fixed EEMC geometry file ecalgeo.g.  Victor checked in a fix on 8/18/2009.  However one additional instance of the buggy line is present.  Also, this bug is intentionally used several places to create a local copy of air called ECAL_CAIR.  These are also corrected.  The geometry file used in this study is attached.
Figure 1 -- Linearity of EEMC sampling fraction for electrons between 0.7 GeV and 250 GeV, at eta = 1.85.  The fit is over the range 10 < E < 200 GeV.  The blue (red) lines indicate the sampling fraction +/- 1% (3%).  The sampling fraction is linear to better than 2% overall, better than 1% for E>7 GeV.