Change BloomFilter implementation for Sparse Joins

build.sbt

@@ -457,7 +457,9 @@ lazy val scioBigQuery: Project = Project(
       // DataFlow testing requires junit and hamcrest
       "org.hamcrest" % "hamcrest-all" % hamcrestVersion % "test,it",
       "com.github.alexarchambault" %% "scalacheck-shapeless_1.13" % scalacheckShapelessVersion % "test,it",
-      "me.lyh" %% "shapeless-datatype-core" % shapelessDatatypeVersion % "test"
+      "me.lyh" %% "shapeless-datatype-core" % shapelessDatatypeVersion % "test",
+      // Our BloomFilters are Algebird Monoids and hence uses tests from Algebird Test
+      "com.twitter" %% "algebird-test" % algebirdVersion % "test"
     )
   )
   .dependsOn(

scio-core/src/main/scala/com/spotify/scio/coders/instances/AlgebirdCoders.scala

@@ -18,11 +18,10 @@
 package com.spotify.scio.coders.instances
 
 import com.spotify.scio.coders.Coder
-import com.twitter.algebird.{BF, Batched, CMS, TopK}
+import com.twitter.algebird.{Batched, CMS, TopK}
 
 trait AlgebirdCoders {
   implicit def cmsCoder[K]: Coder[CMS[K]] = Coder.kryo
-  implicit def bfCoder[K]: Coder[BF[K]] = Coder.kryo
   implicit def topKCoder[K]: Coder[TopK[K]] = Coder.kryo
   implicit def batchedCoder[U]: Coder[Batched[U]] = Coder.kryo
 }