feat(pretrained): integrate pretrained downloader into deepmd-kit

[njzjz-bot]

Summary

This PR integrates the pretrained-dpa functionality directly into deepmd-kit under deepmd/pretrained, while keeping DeepPot usage unchanged.

✅ What is added

• New CLI command:
  • dp pretrained download <MODEL>
• New package folder:
  • deepmd/pretrained/
  • includes: registry.py, download.py, backend.py, entrypoints.py
• Built-in model registry with current models:
  • DPA-3.2-5M
  • DPA-3.1-3M
• Multi-source download strategy:
  • probe multiple URLs in parallel (small range request)
  • rank by latency
  • try fastest first, fallback on failure/timeout/checksum mismatch
• SHA256 verification and atomic .part writes
• .pretrained backend alias support via deepmd/backend/pretrained.py
  • DeepPot("DPA-3.2-5M.pretrained") style remains valid
  • lazy deep-eval adapter to avoid circular import

✅ Command parser integration

• Added pretrained subcommand in deepmd/main.py
• Added dispatch handling in deepmd/entrypoints/main.py

✅ Tests added

• source/tests/common/test_pretrained_parser.py
• source/tests/common/test_pretrained_download.py
• source/tests/common/test_pretrained_backend.py

Notes

• A small fallback metadata file deepmd/__about__.py is included to avoid import failures in source-tree runs when scm-generated version files are absent.

Validation done locally

• python3 -m compileall on changed modules ✅
• python3 -m unittest -q source.tests.common.test_pretrained_parser source.tests.common.test_pretrained_download source.tests.common.test_pretrained_backend ✅
• ruff check on changed files ✅

Authored by OpenClaw (model: custom-chat-jinzhezeng-group/gpt-5.3-codex)

[njzjz-bot]

Good catch. deepmd/__about__.py was a temporary local fallback I added only to make source-tree unittest imports work without the scm-generated version file.

I have removed it now to stay aligned with upstream versioning flow.

Update:

• removed in commit 06c1ff9ad
• no functional change to pretrained integration

Authored by OpenClaw (model: custom-chat-jinzhezeng-group/gpt-5.3-codex)

[njzjz-bot]

Superseded by upstream PR deepmodeling#5277.

[codecov]

Codecov Report

❌ Patch coverage is 58.26087% with 96 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.48%. Comparing base (f959a53) to head (35f384b).

Files with missing lines	Patch %	Lines
deepmd/pretrained/download.py	66.12%	42 Missing ⚠️
deepmd/pretrained/backend.py	26.78%	41 Missing ⚠️
deepmd/pretrained/entrypoints.py	45.45%	6 Missing ⚠️
deepmd/backend/pretrained.py	80.76%	5 Missing ⚠️
deepmd/entrypoints/main.py	33.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            devel     #226      +/-   ##
==========================================
- Coverage   81.57%   81.48%   -0.10%     
==========================================
  Files         750      755       +5     
  Lines       75216    75446     +230     
  Branches     3615     3616       +1     
==========================================
+ Hits        61360    61475     +115     
- Misses      12691    12807     +116     
+ Partials     1165     1164       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.