[Code scan] LAMMPS dump box parser crashes on orthogonal boxes

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

Problem
`dump.get_dumpbox()` always reads a third value from each BOX BOUNDS line as a tilt factor. Standard orthogonal LAMMPS dump boxes use only `lo hi` per axis, so this parser raises `IndexError` for valid dump files.

Code references
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/dump.py#L67
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/dump.py#L72
https://github.com/deepmodeling/dpti/blob/b719828e7eeb571bd26411430197cd74ea78e38c/dpti/lib/dump.py#L75

Reproduction
Parse a dump block with:

```text
ITEM: BOX BOUNDS pp pp pp
0 10
0 10
0 10
```

Expected result
Orthogonal boxes should parse with zero tilt factors instead of crashing.

Start at dpti/lib/dump.py around lines 67, 72, and 75, then reproduce the failure with the orthogonal BOX BOUNDS block from the issue. Verify that standard two-value bounds are accepted and represented with zero tilt factors, without changing handling of tilt values.

Written by the indexing model from the issue text.