[Code scan] HTI range integration ignores the requested scheme

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

Problem
`integrate_range_hti(all_lambda, de, de_err, scheme=...)` accepts a `scheme` argument, but it always calls `integrate_range(..., scheme='s')`. Callers requesting trapezoidal integration still get Simpson-style integration for the main range.

Code references
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/utils.py#L373
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/utils.py#L413
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/utils.py#L414
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/hti.py#L901
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/hti_water.py#L472

Reproduction
A simple example distinguishes the schemes:

```python
import numpy as np
from dpti.lib.utils import integrate_range_hti
x = np.array([0.0, 1.0, 2.0])
y = x * x
err = np.zeros_like(x)
print(integrate_range_hti(x, y, err, scheme='t'))
```

The trapezoidal integral should be 3.0, but the current implementation uses Simpson integration and returns 2.6666667 for the integral.

Expected result
The provided `scheme` should be forwarded to `integrate_range()`.

Start in dpti/lib/utils.py with integrate_range_hti() and compare its call to integrate_range(); review the callers in dpti/hti.py and dpti/hti_water.py for context. Reproduce the provided three-point example with scheme='t' and confirm the result is 3.0 while preserving the existing behavior for other schemes.

Written by the indexing model from the issue text.