Preparations for the release

[rozyczko]

This pull request introduces significant improvements to the handling of zero-variance data points in fitting routines, adds comprehensive documentation for new fitting objectives, and enhances the fitting tutorial to demonstrate backend comparisons. The most important changes include the implementation of Mighell-based objective handling, expanded documentation and user guidance, and updates to the fitting tutorial for clarity and reproducibility.

Fitting objective enhancements:

• Added Mighell-based handling of zero-variance points in fitting routines. Zero-variance data points are now included using a hybrid objective that applies a Mighell substitution for those points, while standard weighted least squares is used for others. The previous behavior of discarding zero-variance points is retained as objective='legacy_mask'. Introduced a new objective parameter on MultiFitter, fit(), and fit_single_data_set_1d().
• Added input validation and utility functions for objective handling, and implemented robust computation of fit metrics for both transformed and classical chi-square calculations.

Documentation improvements:

• Added a new API documentation page for fitting helpers and objective functions, including detailed explanations of Mighell, hybrid, and legacy objective modes, their mathematical basis, and guidance for users.

• Updated the changelog to document the new zero-variance handling and objective parameter.
  Tutorial updates:

• Enhanced the simple fitting tutorial notebook to include both refnx and refl1d backends, showing how to rebuild the model for each and compare results, with clearer output and improved reproducibility.

Other changes:

• Changed the CI workflow to not fail on Codecov upload errors. (.github/workflows/python-ci.yml)

[github-actions]

This pull request does not contain a valid label. Please add one of the following labels: ['chore', 'fix', 'bugfix', 'bug', 'enhancement', 'feature', 'dependencies', 'documentation']