xarray conversion

LawSynth discovery currently accepts a one-dimensional time axis and aligned numeric series. An xarray object with spatial, ensemble, or scenario dimensions must be reduced to an explicitly chosen one-dimensional observation table before fitting.

Choose the coordinate that represents physical time, state the unit and epoch, and document every selection or aggregation across other dimensions. For example, taking a spatial mean changes the target system; it should not be an implicit .values conversion. After selecting the series, convert the time coordinate and each data variable to finite Python lists for Dataset, or write the table to CSV.

Multi-dimensional field discovery and automatic xarray ingestion are not implemented. Keep the xarray preprocessing script in version control and test it against a small fixture with known coordinates.