[Code scan] Conda package omits runtime dependencies while installing with --no-deps

Source: Codex global repository scan of deepmodeling/dpti at commit b719828e7eeb571bd26411430197cd74ea78e38c.
Project: https://github.com/orgs/deepmodeling/projects/27

Problem
The conda recipe installs the package with `pip install --no-deps`, but its run requirements omit several dependencies declared in `setup.py`, including `apache-airflow`, `scipy`, `numpy`, and `pymbar`. The recipe test only imports `dpti`, so missing dependencies for real modules and DAGs are not caught.

Code references
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/setup.py#L12
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/conda/dpti/meta.yaml#L14
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/conda/dpti/meta.yaml#L27
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/conda/dpti/meta.yaml#L33

Reproduction
Build and install the conda package in a clean environment, then run:

```bash
python -c "import dpti.ti; import dpti.dags.dp_ti_gdi"
```

Expected result
The conda runtime dependencies should cover the modules shipped by the package, or optional extras should be split explicitly and tested accordingly.