Pipeline FMAs more deeply by default

Rocket's QoR has improved enough that the FMAs are on the critical
path.  This change seems to keep the integer pipeline's logic
paths balanced with the FPU.

src/main/scala/rocket/fpu.scala

@@ -12,8 +12,8 @@ import cde.{Parameters, Field}
 
 case class FPUConfig(
   divSqrt: Boolean = true,
-  sfmaLatency: Int = 2,
-  dfmaLatency: Int = 3
+  sfmaLatency: Int = 3,
+  dfmaLatency: Int = 4
 )
 
 object FPConstants