Better handling of failing long running index builds in initial sync

When a document is failed to index in initial sync, the build fails and is reattempted 9 times. For example something like the following might be seen:

2015-08-29T08:46:34.920-0600 [rsSync] replSet initial sync exception: 16755 Can't extract geo keys from object, malformed geometry?: { _id: ObjectId('...'), loc: [ x, y ] } 9 attempts remaining

Because this index takes a long time to fail, user intervention could be required (or the node will fall off the oplog) before 9 attempts expire.