compute_sync docstring says optimization is "currently only available for 'accorr'", but WPLI/PLI now have Metal/CUDA/torch backends selected via optimization='auto' (analyses.py:470 is out of date vs sync/wpli.py::_compute_metal + sync/kernels/metal_phase.py::wpli_metal)

[Ramdam17]

The optimization docstring in compute_sync (analyses.py:470) still states "Currently only available for 'accorr'", but the class-based sync API now provides Metal/CUDA/torch backends for WPLI (and PLI) — get_metric('wpli', optimization='auto') resolves to metal on Apple Silicon; please update the docstring.