[SPARK-50515][CORE] Add read-only interface to SparkConf

### What changes were proposed in this pull request?
This PR lifts read-only APIs of `SparkConf` into a new `ReadOnlySparkConf`. `SparkContext` now exposes a new read-only API to the conf through `SparkContext.getReadOnlyConf`, which can be used by clients outside the `spark` package if they require only read-only access. The new API avoids copying the entire (potentially large) conf as in `SparkContext.getConf`. This PR also changes all appropriate call sites to use the new API.

### Why are the changes needed?
Cloning the entire conf adds unnecessary CPU overhead due to copying, and GC overhead due to cleanup. Both affect tail latencies on certain workloads.

### Does this PR introduce _any_ user-facing change?
It adds a new public API `SparkContext.getReadOnlyConf`.

### How was this patch tested?
It is a refactoring PR, so we rely on existing tests.

### Was this patch authored or co-authored using generative AI tooling?
No.

Closes #49100 from pmenon/read-only-confs.

Authored-by: Prashanth Menon <prashanth.menon@databricks.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>

connector/kafka-0-10/src/main/scala/org/apache/spark/streaming/kafka010/DirectKafkaInputDStream.scala

@@ -57,7 +57,7 @@ private[spark] class DirectKafkaInputDStream[K, V](
     ppc: PerPartitionConfig
   ) extends InputDStream[ConsumerRecord[K, V]](_ssc) with Logging with CanCommitOffsets {
 
-  private val initialRate = context.sparkContext.getConf.getLong(
+  private val initialRate = context.sparkContext.getReadOnlyConf.getLong(
     "spark.streaming.backpressure.initialRate", 0)
 
   val executorKafkaParams = {