refactor!: Remove generalized gramians

[ValerianRey]

Closes #690.

Summary

• Engine.compute_gramian now always returns a flat [m, m] PSD matrix where m = output.numel(), regardless of the output shape
• Removed PSDTensor, GeneralizedWeighting, and Flattening — they are no longer needed
• Updated the IWMTL example to use UPGradWeighting directly and reshape the weights before calling backward
• Added a migration guide in CHANGELOG.md

Migration guide

# Before
weighting = Flattening(UPGradWeighting())
gramian = engine.compute_gramian(losses)  # shape: [m1, m2, m2, m1]
weights = weighting(gramian)              # shape: [m1, m2]
losses.backward(weights)

# After
weighting = UPGradWeighting()
gramian = engine.compute_gramian(losses)           # shape: [m1*m2, m1*m2]
weights = weighting(gramian).reshape(losses.shape) # shape: [m1, m2]
losses.backward(weights)

🤖 Generated with Claude Code