docs: describe charge_density as resampled from the raw CHGCAR

README.md

@@ -197,7 +197,7 @@ density = np.asarray(row["charge_density"]).reshape(row["charge_density_grid_sha
 ## Notes
 
 - **Charge density compression**: The `compressed_*` columns are lossy-compressed using [pyrho](https://github.com/materialsproject/pyrho)'s `lossy_smooth_compression`, which smooths the density with a Gaussian (`smear_std=0.2`) before resampling it. This is suitable for qualitative analysis and ML feature extraction but not for quantitative integration (e.g. re-running Bader). The raw CHGCAR files are available on request.
-- **Unsmoothed charge density**: `charge_density` resamples the archived CHGCAR onto the same grid by Fourier interpolation, without smoothing, and is stored as float32 rather than JSON text. It differs from `compressed_charge_density` by about 15% (normalized mean absolute error). `reshape(charge_density_grid_shape)` gives the same [x, y, z] indexing as `np.array(json.loads(row["compressed_charge_density"]))`. Fourier interpolation can leave small negative values near the atoms, especially on the 15×15×15 grid. The AECCAR columns are only available smoothed.
+- **Unsmoothed charge density**: `charge_density` resamples the raw CHGCAR onto the same grid by Fourier interpolation, without smoothing, and is stored as float32 rather than JSON text. It differs from `compressed_charge_density` by about 15% (normalized mean absolute error). `reshape(charge_density_grid_shape)` gives the same [x, y, z] indexing as `np.array(json.loads(row["compressed_charge_density"]))`. Fourier interpolation can leave small negative values near the atoms, especially on the 15×15×15 grid. The AECCAR columns are only available smoothed.
 - **Bader coverage**: ~13.8% of materials lack Bader charges because their AECCAR files are absent from the source S3 data. This is consistent across all three source databases and reflects an upstream calculation issue.
 - **DDEC6 charges**: `ddec6_charges` is null in this release. A follow-up patch run is planned.
 - **r2SCAN functional**: All calculations use the r2SCAN meta-GGA functional. MP2020 energy corrections are PBE-specific and do not apply; `energy_corrected` equals `energy` for all rows.