feat: prepare for open source release with KPS v1.2 support

[zhexuany]

Overview

  This is the initial open-source release of Robocodec v0.1.0. This PR adds comprehensive KPS (Kupas) dataset format support, complete pipeline architecture, CLI tools, and all necessary open-source
  infrastructure.

  📊 Statistics

  - 197 files changed
  - 44,363 insertions
  - 5,903 deletions
  - 782 tests passing

  🎯 Key Features

  1. KPS Dataset Format (New)

  - V1.2 Specification: Compliant directory structure with Robot-EndEffector-Scene naming
  - Fluent API: Type-safe builder for conversions
  - Statistics Tracking: Automatic directory naming with actual stats
  - Time Alignment: Linear, Nearest Neighbor, Hold Last strategies
  - Multi-format Output: HDF5, Parquet, MP4
  - Audio Support: WAV file output
  - Original Data: proprio_stats_original.hdf5 for unaligned data

  2. Pipeline Architecture
  ┌───────────────┬────────┬────────────┬────────────────────┐
  │   Pipeline    │ Stages │ Throughput │      Use Case      │
  ├───────────────┼────────┼────────────┼────────────────────┤
  │ Standard      │ 4      │ ~200 MB/s  │ Balanced           │
  ├───────────────┼────────┼────────────┼────────────────────┤
  │ HyperPipeline │ 7      │ ~1800 MB/s │ Max throughput     │
  ├───────────────┼────────┼────────────┼────────────────────┤
  │ KPS Pipeline  │ 5+     │ Varies     │ Dataset conversion │
  └───────────────┴────────┴────────────┴────────────────────┘
  3. CLI Tools

  robocodec convert    # Format conversion
  robocodec inspect    # File inspection
  robocodec extract    # Data extraction
  robocodec schema     # Schema utilities
  robocodec search     # Data search

  4. Open Source Infrastructure

  - ✅ MulanPSL-2.0 license (Chinese OSI-approved license)
  - ✅ Bilingual documentation (English & Chinese)
  - ✅ CI/CD workflows (test + publish)
  - ✅ Issue/PR templates
  - ✅ Contributing guidelines
  - ✅ Security policy
  - ✅ Code of conduct

  📝 Documentation

  - README.md / README_zh.md - Project overview
  - ARCHITECTURE.md - High-level architecture
  - PIPELINE.md - Pipeline details
  - MEMORY.md - Memory management
  - CONTRIBUTING.md - Contribution guide
  - examples/kps/ - KPS conversion examples

  🔧 Feature Flags
  ┌─────────────┬─────────────────────────┐
  │    Flag     │       Description       │
  ├─────────────┼─────────────────────────┤
  │ kps-hdf5    │ HDF5 dataset support    │
  ├─────────────┼─────────────────────────┤
  │ kps-parquet │ Parquet dataset support │
  ├─────────────┼─────────────────────────┤
  │ kps-depth   │ Depth video support     │
  ├─────────────┼─────────────────────────┤
  │ kps-all     │ All KPS features        │
  └─────────────┴─────────────────────────┘
  📦 Examples

  // KPS conversion with fluent API
  use robocodec::pipeline::kps::KpsConverter;

  let report = KpsConverter::new("input.mcap", "output_dir")
      .config("config.toml")
      .v12_delivery()
      .robot("Kuavo4Pro")
      .end_effector("Dexhand")
      .scene("Housekeeper")
      .sub_scene("Kitchen")
      .task("Dispose_of_takeout_containers")
      .with_statistics()
      .run()?;

  ✅ Test Results

  test result: ok. 782 passed; 0 failed

  📄 License

  This project is licensed under the MulanPSL v2 - see LICENSE for details.



<!-- CURSOR_SUMMARY -->
---

> [!NOTE]
> Initial open-source setup, repo restructure, and CI/CD.
> 
> - **Licensing & compliance**: Add MulanPSL-2.0 `LICENSE`, `NOTICE`, REUSE config, and Code of Conduct
> - **Docs**: New bilingual README, `ARCHITECTURE.md`, `PIPELINE.md` refs, `MEMORY.md`, ROADMAP, SECURITY, CONTRIBUTING, CLAUDE guidance
> - **CI/CD**: GitHub Actions for lint/tests (Rust/Python), type-checking, coverage (Codecov), and publish to PyPI/crates.io
> - **Templates & tooling**: Issue/PR templates, Makefile, codecov config, Serena/Claude settings
> - **Workspace & build**: Convert to Cargo workspace (`roboflow` + `robocodec`), update `Cargo.toml` metadata, dependencies, and features (`python`, `kps-*`, `profiling`, `jemalloc`, `gpu`)
> - **CLIs & bench**: Add binaries `convert`, `extract`, `inspect`, `schema`, `search`; add `benches/profiler` with flamegraph support; remove `extract_sample`
> - **Housekeeping**: Expanded `.gitignore` and initial CHANGELOG
> 
> <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit 6216f603e8b92a15d6f02dc092a63c1a349182e3. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->

[05F3759DF]

1. As MulanPSL2 says, the whole licence should be contained(including Chinese version). The Chinese verison should prevail if there are conflicts.
2. The newest version of Contributor Covenant is 3.0, which boosts cultural sensitivity, restorative justice, and customization.
3. MCAP maybe is for Modular Container for Associated Protocols, not Mission data CAPture.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}

[cursor]

Profile command fails with filename-only output path

Medium Severity

When --profile-output is a filename without a directory component (e.g., myprofile), parent() returns Some(Path::new("")) (empty path) rather than None. The unwrap_or(Path::new(".")) doesn't trigger because Some was returned, so profile_dir becomes an empty string. Then create_dir_all("") fails with a confusing error about creating an empty-named directory, rather than correctly defaulting to the current directory.