KfComponentsHolder cleanup

[VinInn]

Cleanup of KfComponentsHolder.
It became a bit more urgent due to changes introduced by thread-safety.
Restore the timing performance of the Chi2MeasurementEstimator (still pretty deceiving though).

No regression expected, no regression observed

[cmsbuild]

A new Pull Request was created by @VinInn (Vincenzo Innocente) for CMSSW_7_2_X.

KfComponentsHolder cleanup

It involves the following packages:

Alignment/ReferenceTrajectories
DataFormats/Math
DataFormats/TrackerRecHit2D
DataFormats/TrackingRecHit
TrackingTools/GsfTracking
TrackingTools/KalmanUpdators

@nclopezo, @cmsbuild, @diguida, @rcastello, @StoyanStoynev, @slava77, @Degano can you please review it and eventually sign? Thanks.
@ghellwig, @frmeier, @GiacomoSguazzoni, @rovere, @gpetruc, @cerati, @mmusich, @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.

[cmsbuild]

+1
Tested at: 2f8b670
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-4447/213/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-4447/213/summary.html

[StoyanStoynev]

Typo, should be as before "size_ = 0;".

[VinInn]

oops...
anyhow it compiles only for debug, so does not affect testing and production,
will be fixed

[StoyanStoynev]

+1
Tested 2f8b670 on top of CMSSW_7_2_X_2014-06-27-0200.
Used extended runMatrix tests. No regressions observed and no other issues. Within few %, I see no difference in timing with wf 202 (ttbar with PU) and there is upto10 % positive effect on tracking modules in wf 16 and less clear in 17 (high and lower pt electron guns). For instance:
wf 202 (200 events):
electronGsfTracks 190.725 ms/ev -> 175.76 ms/ev
conversionTrackCandidates 98.4586 ms/ev -> 93.1699 ms/ev
(references: clusterSummaryProducer 63.7899 ms/ev -> 57.6302 ms/ev; siPixelClusterShapeCache 2.57173 ms/ev -> 2.3351 ms/ev)

wf 16 (2k events):
electronGsfTracks 744.966 ms/ev -> 631.855 ms/ev
conversionTrackCandidates 2179.84 ms/ev -> 1742.92 ms/ev
(references: clusterSummaryProducer 13.3835 ms/ev -> 11.568 ms/ev; siPixelClusters 17.4552 ms/ev -> 15.0882 ms/ev)

wf 17 (2k events):
electronGsfTracks 690.541 ms/ev -> 581.271 ms/ev
conversionTrackCandidates 1062.49 ms/ev -> 829.525 ms/ev
(references: clusterSummaryProducer 6.20514 ms/ev -> 5.02205 ms/ev; siPixelClusters 13.0698 ms/ev -> 9.72796 ms/ev; others also point to ~20% diff)

The bulk of the code changes consists of code restructuring. The non-trivial part deals with time optimizations (including usage of different objects).

[StoyanStoynev]

... and a typo in the code should be fixed (no effect on default running).

[cmsbuild]

+1
Tested at: 63ce78b
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-4447/259/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-4447/259/summary.html

[StoyanStoynev]

Working on it. I'll compare the main features I observed with a previous commit using wf 202.
Jenkins are fine EXCEPT the many new warnings (compared with the previous tests) in the Unit tests.
These are probably not from this PR but one of the warnings (type) I mentioned in #4435:

%MSG-w TwoTrackMinimumDistance: PFTauPrimaryVertexProducer:hpsPFTauPrimaryVertexProducerPFlow

The others are for CondDB.

[StoyanStoynev]

+1
Retested: 63ce78b on top of CMSSW_7_2_X_2014-07-04-0200. Performed extended tests with wf 202 and 16 (as before).
No issues, see the review above.
This time the machine was overloaded most of the time (that is why it took more time to complete as well) - the timing tests are less reliable but don't suggest excessive behavior just more jitter.
The code is mostly unchanged from the previously tested version though it is not identical (there are also changes to resolve a conflict, see above).

[diguida]

+1

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next CMSSW_7_2_X IBs unless changes (tests are also fine).