Replace SMatrixNoInit

[wmtan]

ROOT6 has replaced SMatrixNoInit() with SMatrixIdentity().
This pull request adapts CMSSW to this change, and fixes most, perhaps all, of the compilation errors in the latest ROOT6 IB. Please merge ASAP.

[cmsbuild]

A new Pull Request was created by @wmtan for CMSSW_7_2_ROOT6_X.

Replace SMatrixNoInit

It involves the following packages:

DataFormats/Math
RecoVertex/KinematicFit
TrackingTools/AnalyticalJacobians
TrackingTools/GsfTools
TrackingTools/TrajectoryParametrization

@nclopezo, @cmsbuild, @thspeer, @StoyanStoynev, @slava77, @Degano can you please review it and eventually sign? Thanks.
@gpetruc, @cerati, @GiacomoSguazzoni, @rovere, @venturia this is something you requested to watch as well.
You can sign-off by replying to this message having '+1' in the first line of your reply.
You can reject by replying to this message having '-1' in the first line of your reply.

[ktf]

SMatrixNoInit was actually introduced in Root5 by @VinInn, to avoid unneeded initializations and speedup the Reco by ~3%. Since the patch still applies for root6, I've simply forward ported the patch. CMSSW_7_2_ROOT6_X_2014-31-05-0200 should have the change. I'll follow up with the root people to see if we can get it included in the first convenient patch release of root6.

[VinInn]

please close this PR.
I am sorry for the confusion created in the root6 branch by the changes in the main trench

[ktf]

1. The CMSSW_7_2_ROOT6_X_2014-31-05-0200 build failed, if I interpret
   Jenkins correctly.

No, it did not. I stopped it and restarted with the new patch applied.
It's still building.

2. I need a working IB by Monday or i am dead in the water. If the
   problem with the build has anything to do with the SMatrixNoInit patch
   and it cannot be fixed by Monday, Fermilab time, I would request that
   this pull request be merged as a temporary workaround so I can get a
   good IB. We can easily back it out later when SMatrixNoInit is
   working.

I'll do my best to make sure you have something working by monday at the
latest, but hopefully in just a few hours.

3. I need a copy of the root6 patch adding SMatrixNoInit()

cms-sw/root@a792dbb

[wmtan]

The new patch reproducibly causes segfaults in genreflex during building over 30 packages.
Unless this can be fixed by Monday morning, Fermilab time, I request the backout of the patch and the merging of this pull request ASAP. I need a working IB. The 3% RECO performance improvement can wait until ROOT6 is functioning.

[ktf]

Ok, I reverted the patch and merged this (ROOT6) only. I'll check tomorrow why there were crashes.

[wmtan]

Thanks. The crashes are reproducible. Unless they are solved first by someone else, I will show them to Philippe soon, and we can discuss how to proceed. In the meantime, this issue will not interfere with other ROOT6 work.

[wmtan]

The patch was reverted, but the changes for this pull request are not in the latest IB, CMSSW_7_2_ROOT6_X_2014-06-01-2200. I hope they will be in the next IB.

[ktf]

Yes, 0200 should be the first one to have them.