Add Knitro solver support

[FabianHofmann]

Summary

This PR adds support for the Artelys Knitro solver to linopy.

Changes

• Added automatic detection of the knitro Python package to the available solvers list
• Implemented Knitro solver class with support for solving MPS/LP files
• Registered Knitro capabilities in the solver registry (quadratic objectives, LP file names, no solution file needed)
• Added tests for the Knitro solver (properly skip when package not installed)
• Mapped Knitro status codes to linopy's Status system

Implementation Details

The Knitro solver implementation follows the existing patterns in linopy:

• File-based solver using MPS/LP files (similar to CPLEX, COPT)
• Proper status code mapping for optimization results
• Solution and dual value extraction using Knitro's Python API
• Support for solver options and logging

Testing

• Added unit tests that verify Knitro functionality
• Tests properly skip when Knitro package is not installed
• All existing tests continue to pass

Notes

• Direct API support (solve_problem_from_model) is not implemented in this initial version but can be added later
• Knitro is a powerful nonlinear optimization solver that also handles linear and quadratic problems efficiently
• Users need to install the knitro package separately and have a valid license to use this solver

Closes #[issue_number] (if applicable)