refactor(box): remove legacy in-process runtime code and clean up smells

After the architecture settled on always using an independent Box Runtime
service, several pieces of compatibility code and design shortcuts were
left behind. This commit cleans them up:

- Remove `LocalBoxRuntimeClient` and `create_box_runtime_client` from
  production code (moved to test-only helper).
- Remove unused `_clip_bytes` method from backend.
- Remove `__langbot_session_placeholder__` hack by making `BoxSpec.cmd`
  default to empty and validating non-empty only in `runtime.execute()`.
- Extract `get_box_config()` helper to eliminate 5× duplicated config
  access boilerplate.
- Remove `session_id`/`host_path`/`host_path_mode` from the LLM-facing
  tool schema to enforce request-scoped session isolation.
- Fix dual shutdown path: `NativeToolLoader.shutdown()` no longer calls
  `box_service.shutdown()` (handled by `Application.dispose()`).
- Simplify `_assert_session_compatible` with a loop.
- Inline client creation in `BoxRuntimeConnector`.
- Remove redundant `BOX__RUNTIME_URL` env var from docker-compose
  (auto-detected by code).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

tests/unit_tests/box/test_backend_clip.py

@@ -5,33 +5,34 @@ import pytest
 from langbot.pkg.box.backend import CLISandboxBackend, _MAX_RAW_OUTPUT_BYTES
 
 
-class TestClipBytes:
+class TestClipCapturedBytes:
     def test_within_limit_unchanged(self):
         data = b'hello world'
-        result = CLISandboxBackend._clip_bytes(data, limit=1024)
+        result = CLISandboxBackend._clip_captured_bytes(data, total_size=len(data), limit=1024)
         assert result == 'hello world'
 
     def test_exceeding_limit_clips_and_appends_notice(self):
-        data = b'A' * 200
-        result = CLISandboxBackend._clip_bytes(data, limit=100)
+        captured = b'A' * 100
+        total_size = 200
+        result = CLISandboxBackend._clip_captured_bytes(captured, total_size=total_size, limit=100)
         assert result.startswith('A' * 100)
         assert 'raw output clipped at 100 bytes' in result
         assert '100 bytes discarded' in result
 
     def test_exact_limit_not_clipped(self):
         data = b'B' * 100
-        result = CLISandboxBackend._clip_bytes(data, limit=100)
+        result = CLISandboxBackend._clip_captured_bytes(data, total_size=100, limit=100)
         assert result == 'B' * 100
         assert 'clipped' not in result
 
     def test_default_limit_is_module_constant(self):
         data = b'x' * 10
-        result = CLISandboxBackend._clip_bytes(data)
+        result = CLISandboxBackend._clip_captured_bytes(data, total_size=10)
         assert result == 'x' * 10
         assert _MAX_RAW_OUTPUT_BYTES == 1_048_576
 
     def test_invalid_utf8_replaced(self):
         data = b'ok\xff\xfetail'
-        result = CLISandboxBackend._clip_bytes(data, limit=1024)
+        result = CLISandboxBackend._clip_captured_bytes(data, total_size=len(data), limit=1024)
         assert 'ok' in result
         assert 'tail' in result