v1.0.2

Improvements

• Use diffcp.solve_only_batch when gradients aren't needed (#204)
• Centralize CSC→CSR conversion outside solver interfaces (#203)
• Enable DPP with quad_form (#208)
• Add warm-start support for Moreau solver (#209)
• Add Moreau solver to CI (#206)

Examples & Docs

• Added optimal transport example (#43)
• Fix PyTorch tutorial: zero Xtrain.grad before using it (#102)