[SPARK-58021][CONNECT] Warm local server pool members before use

[ericm-db]

What changes were proposed in this pull request?

This is layer 6 of the local Connect pool stack: #57684 -> #57685 -> #57686 -> #57687 -> #57102 -> this PR. Until the lower layers merge, GitHub shows their cumulative diff; the new review unit here is commit 275235cd137.

This layer adds optional JIT warmup for newly booted pool members:

• attendants launch a bounded child process against each published server;
• fixed synthetic queries warm JVM JIT and Spark codegen paths without creating persistent catalog state;
• warmup stops immediately if the member is claimed and is killed after its timeout;
• SPARK_LOCAL_CONNECT_POOL_WARMUP=0 disables it;
• the internal --warm entry point runs the query set; and
• the pool documentation explains the latency and memory trade-off.

Why are the changes needed?

A booted server avoids JVM and SparkContext startup, but its first real query can still pay JIT and code-generation costs. Keeping warmup separate from acquisition makes this optimization and its subprocess failure boundary optional and independently reviewable.

Does this PR introduce any user-facing change?

Yes, only for the experimental pool opt-in introduced by #57102. Pool members warm themselves by default before use. Set SPARK_LOCAL_CONNECT_POOL_WARMUP=0 to retain the booted-server behavior without synthetic queries. Behavior is unchanged when the pool opt-in is disabled.

How was this patch tested?

Added focused tests for environment parsing, termination of a warmup child after a member is claimed, and execution plus cleanup of the fixed query set. The full pool suite also covers sequential and concurrent clients against real local Connect servers.

python -m unittest -v pyspark.sql.tests.connect.test_connect_local_server_pool

All 22 tests passed. Ruff check, Ruff format check, and git diff --check also passed.

Was this patch authored or co-authored using generative AI tooling?

Generated-by: Claude Code (Fable 5) and OpenAI Codex (GPT-5)

[uros-b]

Thank you @ericm-db! cc @HyukjinKwon