row-count: require f:c, not file:size

I tried using the empty List() and got a test failure, so it seems like
we do need to specific *some* field here.

file:size gets populated by the extraction job, not the backfill job, so
I had been miscounting table sizes (counting only the number of GROBID
extracted items, not rows in the table).

TODO: count on key or no column, not f:c

1 files changed, 1 insertions, 1 deletions

diff --git a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
index 4c3de33..5c7954a 100644
--- a/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
+++ b/scalding/src/main/scala/sandcrawler/HBaseRowCountJob.scala
@@ -30,7 +30,7 @@ object HBaseRowCountJob {
     HBaseBuilder.build(
       hbaseTable,
       zookeeperHosts,
-      List("file:size"),
+      List("f:c"),
       SourceMode.SCAN_ALL)
   }
 }