Added sourceSelector functionally to matchOptimisticB.py and modified it's unittests(testMatchOptimisticB, testAstrometry, testJoinMatchListWithCatalog) to contain a "deblend_nChild" column. The code currently passes all unittests and properly runs on a processCcd run of DECam data.

matchSourceSelector is currently used for replacement of the isUsable flag from the old SourceInfo. The other flag, isGood which is used after matches have been found, is currently hard coded. This is used as the current API of sourceSelector can not handle a matchList object.

The implemented changes can be found at https://github.com/lsst/meas_astrom/tree/u/morriscb/sourceSelector

Please rebase squash/flatten before merging, and link the community posting you made about this topic to this ticket (and/or vice-versa).

Added ticket branch to pipe_tasks regarding compliance with the new sourceSelectors. Pull request can be found here: https://github.com/lsst/pipe_tasks/pull/92

All tickets currently pass Jenkins py2, waiting on py3 and ci_hsc.

Added compliance with the new sourceSelector used in matchOptimisticB to obs_subaru. The pull request can be found here: https://github.com/lsst/obs_subaru/pull/50

Instead of using astrometrySourceSelector a new sourceSelector, matcherSourceSelector to replace the isUsable criteria in the old SourceInfo class. astrometrySourceSelector can not be used in matchOptimisticB currently as the source selector can not be applied to the catalog like object, matchList in the code. A discussion on what is needed to change in the sourceSelector class can be found here https://community.lsst.org/t/should-sourceselectors-work-for-arbitrary-collections-of-record-objects/1549