docs: add missing examples to Python and Node bindings

[chaliy]

Context

Node has 5+ runnable examples covering diverse use cases. Python has only 2. Each binding should have equivalent examples demonstrating the same patterns, adapted to language idioms.

This is part of the Python ↔ Node binding parity effort (Phase 4 — Documentation Alignment).

Current state

Example pattern	Python (examples/)	Node (examples/)
Basic bash usage	✅ bash_basics.py	✅ bash_basics.ts
Data pipeline	❌	✅ data_pipeline.ts
LLM tool usage	❌	✅ llm_tool.ts
K8s orchestrator (ScriptedTool)	✅ k8s_orchestrator.py	❌
LangChain integration	❌	✅ langchain_integration.ts

What to implement

Add to Python (crates/bashkit-python/examples/)

data_pipeline.py — Data processing pipeline example

• Read/generate sample data (CSV or JSON)
• Use bash pipelines (sort, uniq, awk, jq) to transform data
• Demonstrate VFS for intermediate files
• Show both sync and async execution
• Mirror the patterns in Node's data_pipeline.ts

llm_tool.py — LLM tool usage example

• Create a BashTool instance
• Show tool metadata (description(), input_schema(), output_schema())
• Demonstrate the execute → result → parse cycle an LLM agent would use
• Show system_prompt() for LLM context
• Mirror the patterns in Node's llm_tool.ts

Add to Node (crates/bashkit-js/examples/)

k8s_orchestrator.ts — ScriptedTool orchestration example

• Create a ScriptedTool with mock K8s API tools (get-pods, get-deployments, scale, etc.)
• Register 5+ tools with callbacks and schemas
• Execute a bash script that orchestrates multiple tools via pipes and jq
• Mirror the patterns in Python's k8s_orchestrator.py

Acceptance criteria

• data_pipeline.py created and runs successfully: python examples/data_pipeline.py
• llm_tool.py created and runs successfully: python examples/llm_tool.py
• k8s_orchestrator.ts created and runs successfully (via ts-node or compiled)
• Each example has a header comment explaining what it demonstrates
• Examples are self-contained (no external dependencies beyond bashkit)
• Examples use realistic, non-trivial patterns (not just "hello world")
• Python examples follow existing style in bash_basics.py (print output, show both sync/async)
• Node example follows existing style in other TS examples
• CI smoke tests updated to run the new examples:
  • Python: add data_pipeline.py and llm_tool.py to .github/workflows/python.yml examples job
  • Node: add k8s_orchestrator.ts to .github/workflows/js.yml examples job