fix: probe CollectionConverters with a call to detect Scala version correctly

On Scala 2.12, jvm.scala.jdk.javaapi.CollectionConverters resolves to a
JavaPackage placeholder rather than raising an exception, so the previous
try/except always selected the jdk path and broke Spark 3.x.
Verify usability by actually calling asScala() as a one-time probe.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: m-aciek

pydeequ/scala_utils.py

@@ -81,7 +81,12 @@ def _get_converters(jvm):
     key = id(jvm)
     if key not in _jvm_converters_cache:
         try:
-            _jvm_converters_cache[key] = ("jdk", jvm.scala.jdk.javaapi.CollectionConverters)
+            converters = jvm.scala.jdk.javaapi.CollectionConverters
+            # On Scala 2.12, the path resolves to a JavaPackage placeholder (no class
+            # exists), so attribute access succeeds but any method call raises TypeError.
+            # Probe with an actual call to confirm the class is genuinely usable.
+            converters.asScala(jvm.java.util.ArrayList())
+            _jvm_converters_cache[key] = ("jdk", converters)
         except Exception:
             _jvm_converters_cache[key] = ("legacy", jvm.scala.collection.JavaConverters)
     return _jvm_converters_cache[key]