[Halld-offline] Changes to track finding in the CDC

[Simon Taylor]

Hi.

I've implemented a preliminary fix to the "bogus track candidate" 
problem I brought up at the previous tracking meeting.  (For those that 
were not present,  the default cdc track finder frequently gets confused 
when dealing with tracks crossing each other or decaying particles such 
as Lambdas causing V-like patterns , generating many more track 
candidates than actual tracks).    I have updated the wiki page 
(https://halldweb1.jlab.org/wiki/index.php/Problems_with_track_finding_in_CDC) 
where I first highlighted the problem with an explanation of the 
technique and some reconstructed lambda and kzero events.   I have 
created a new command-line flag to turn on/off this behavior:  set 
-PTRKFIND:FILTER_SEEDS=1 to enable the new algorithm.  By default for 
now it is off.

Simon