sbt 2.x: run integration tests sequentially to avoid Postgres connection exhaustion

Root cause of the sbt 2.0 test failures: 'FATAL: remaining connection slots are reserved'.
Each integration spec opens Hikari pools (users+discussions x connectionPool) against the one
shared Postgres; sbt 2.x runs test classes in the forked JVM more concurrently than sbt 1.x,
blowing past Postgres's connection limit -> flyway migrate / queries fail with
'ResultSet exhausted'. Set Test/parallelExecution=false + testForkedParallel=false for the
integration-test modules (Tags.limit(Test,1) alone only serialized across modules).

Author: MateuszKubuszok

branchtalk.sbt

@@ -131,7 +131,11 @@ val tests = Seq(
 ) ++ scalafmtConfigSettings(Test)
 
 val integrationTests = tests ++ Seq(
-  Test / fork := true
+  Test / fork := true,
+  // Integration tests each open Hikari pools against the one shared Postgres; sbt 2.x runs test classes in a forked JVM
+  // more concurrently than sbt 1.x, exhausting Postgres connection slots. Run them sequentially.
+  Test / parallelExecution := false,
+  Test / testForkedParallel := false
 )
 
 def customPredef(imports: String*): Def.Setting[Task[Seq[String]]] =