chore: fix typing with new highspy version

[FBumann]

mypy fix due to tightened stubs in highspy 1.15

Note

This change was generated by AI.

The direct-solve HiGHS path built the integrality array as int32 and passed it to changeColsIntegrality, whose var_types argument expects a uint8 array — matching HiGHS's uint8-backed HighsVarType enum (Continuous=0, Integer=1, SemiContinuous=2, …).

Recent highspy (1.13.1) tightened its type stubs to require uint8 here, which surfaces as a mypy error:

linopy/solvers.py: error: Argument 3 to "changeColsIntegrality" ...
  has incompatible type "ndarray[..., dtype[signedinteger[_32Bit]]]";
  expected "ndarray[..., dtype[unsignedinteger[_8Bit]]]"

The persistent in-place update path (_apply_var_types) already builds this array as uint8; this aligns the direct-solve path. uint8 is accepted at runtime across highspy versions, so this is a safe correctness fix, not a version workaround.

[codspeed-hq]

Merging this PR will not alter performance

✅ 138 untouched benchmarks
⏩ 138 skipped benchmarks¹

---

_{Comparing fluxopt:fix/highs-integrality-uint8 (d4dcafe) with master (2d47d4a)}

Footnotes

1. 138 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[FBumann]

I canceled pypsa run to save ci minutes