feat: env-configurable spark-submit binary name

Hardcoding 'spark-submit' as cmd[0] in build_spark_submit_command
breaks for hosts where:
  - both Spark 1.x and 2.x/3.x are installed and 'spark-submit' resolves
    to the wrong one (use 'spark2-submit' or 'spark3-submit' explicitly)
  - the user wants to launch via the PySpark entrypoint ('pyspark')
  - a custom wrapper script sits on PATH (e.g. a credentials-injecting
    'spark-submit-wrapper')

New env var SPARK_EXECUTOR_SPARK_SUBMIT_BIN. Default is 'spark-submit'
(preserves the current behavior for everyone). Override in .env /
docker-compose.yml to change.

common/config.py:
  - new Settings.spark_submit_bin field
  - env-var resolution in from_env() with default 'spark-submit'
  - included in reload() so tests work

spark_executor/core/spark_submit.py:
  - cmd[0] reads settings.spark_submit_bin (was hardcoded 'spark-submit')

.env.example: new section with the override and example values.
docker-compose.yml: forwards the var with the standard 'spark-submit'
default.

Tests: 2 new (settings.spark_submit_bin='spark2-submit', ='pyspark')
plus existing tests updated to use the settings-restore fixture so
mutations don't leak between tests.

165/163 still pass.

tests/unit/test_spark_submit.py

@@ -3,6 +3,7 @@ from unittest.mock import MagicMock, patch
 
 import pytest
 
+from common import config
 from spark_executor.core.spark_submit import (
     SparkSubmitError,
     build_spark_submit_command,
@@ -10,6 +11,25 @@ from spark_executor.core.spark_submit import (
 )
 
 
+@pytest.fixture(autouse=True)
+def _restore_settings():
+    snapshot = config.Settings(
+        data_dir=config.settings.data_dir,
+        jobs_dir=config.settings.jobs_dir,
+        log_dir=config.settings.log_dir,
+        yarn_resource_manager_url=config.settings.yarn_resource_manager_url,
+        log_level=config.settings.log_level,
+        spark_submit_bin=config.settings.spark_submit_bin,
+    )
+    yield
+    config.settings.data_dir = snapshot.data_dir
+    config.settings.jobs_dir = snapshot.jobs_dir
+    config.settings.log_dir = snapshot.log_dir
+    config.settings.yarn_resource_manager_url = snapshot.yarn_resource_manager_url
+    config.settings.log_level = snapshot.log_level
+    config.settings.spark_submit_bin = snapshot.spark_submit_bin
+
+
 def test_build_command_uses_provided_master_and_deploy_mode():
     cmd = build_spark_submit_command(
         master="yarn",
@@ -20,7 +40,7 @@ def test_build_command_uses_provided_master_and_deploy_mode():
         executor_cores=2,
         num_executors=2,
     )
-    assert cmd[:2] == ["spark-submit", "--master"]
+    assert cmd[:2] == ["spark-submit", "--master"]  # default binary
     assert "yarn" in cmd
     assert "cluster" in cmd
     assert "--queue" in cmd and "default" in cmd
@@ -63,6 +83,37 @@ def test_build_command_appends_spark_conf_entries():
     assert cmd[-1] == "/tmp/j.py"
 
 
+def test_build_command_uses_settings_spark_submit_bin():
+    """The Spark CLI binary is env-configurable via settings.spark_submit_bin
+    (env var SPARK_EXECUTOR_SPARK_SUBMIT_BIN). Default is 'spark-submit';
+    override to 'spark2-submit', 'pyspark', etc. for mixed-version hosts."""
+    config.settings.spark_submit_bin = "spark2-submit"
+    cmd = build_spark_submit_command(
+        master="yarn",
+        deploy_mode="cluster",
+        script_path="/tmp/j.py",
+        queue="default",
+        executor_memory="4G",
+        executor_cores=2,
+        num_executors=2,
+    )
+    assert cmd[0] == "spark2-submit"
+
+
+def test_build_command_supports_pyspark_binary():
+    config.settings.spark_submit_bin = "pyspark"
+    cmd = build_spark_submit_command(
+        master="yarn",
+        deploy_mode="cluster",
+        script_path="/tmp/j.py",
+        queue="default",
+        executor_memory="4G",
+        executor_cores=2,
+        num_executors=2,
+    )
+    assert cmd[0] == "pyspark"
+
+
 def test_run_spark_submit_returns_completed_process(monkeypatch):
     fake = MagicMock()
     fake.returncode = 0