Refactor correlation logic for flexible 180° projection selection

[ashleymeigh2]

Issue

Closes #2685

Description

• Removed fallback logic to automatically generate or use a 180° projection.
• Introduced support for user-specified projection indices for CoR correlation.
• Added validation to ensure selected projections are ~180° apart.
• Cleaned up old proj180deg assumptions and replaced with generalised correlation method.
• Updated reconstruction workflow to respect manual projection selection for correlation.

Developer Testing

• I have verified unit tests pass locally: python -m pytest -vs
• Updated or fixed failing unit tests related to correlation logic

Acceptance Criteria and Reviewer Testing

• Unit tests pass locally: python -m pytest -vs
• Selecting two projections and running "Find CoR & Tilt" works with ~180° apart images
• Dataset does not auto-generate or fallback to a 180° projection
• Old behaviour relying on proj180deg is not triggered
• Reconstruct workflow integrates correctly with the updated correlation logic

Documentation and Additional Notes

• Release Notes have been updated (if feature-level change)
• Sphinx documentation has been updated (if API/docs exposed to users)

[samtygier-stfc]

Just had a quick look at this. Is it still possible to use the 180 project if you have loaded one? IMAT do usually record one, so in that case it should still be possible to use it. (This should have been clearer on the issue)