Missing geometry from cells_dataframe_to_geodataframe with non-sequential index

[cas--]

Encountered a rather puzzling bug where a non-sequential index for a dataframe causes a problem when converting cells to geodataframe resulting in missing geometry for most of the cells.

uvx --with h3ronpy[pandas] python

from h3ronpy.pandas.vector import cells_dataframe_to_geodataframe
import pandas as pd
df = pd.DataFrame({'cell': {2: 630950036828342783, 19: 630950036828350975, 38: 630950036828390911, 55: 630950036828391423}})
cells_dataframe_to_geodataframe(df)

                  cell                                           geometry
2   630950036828342783  POLYGON ((-3.42058 50.63746, -3.42072 50.63743...
19  630950036828350975                                               None
38  630950036828390911                                               None
55  630950036828391423                                               None

Resetting the index is a workaround for the issue

cells_dataframe_to_geodataframe(df.reset_index(drop=True))

0  630950036828342783  POLYGON ((-3.41817 50.6377, -3.41831 50.63767,...
1  630950036828350975  POLYGON ((-3.41884 50.63765, -3.41898 50.63762...
2  630950036828390911  POLYGON ((-3.42058 50.63746, -3.42072 50.63743...
3  630950036828391423  POLYGON ((-3.42048 50.6373, -3.42062 50.63727,...

[cas--]

This likely should be a separate feature request but I did want to ask how easy the geometry decimal precision can be modified since when comparing output with duckdb h3 h3_cell_to_boundary_wkt I noticed their WKT output is only 6 decimal places (seems recommended for mapping) compared to 12 for this library:

>>> str(cells_dataframe_to_geodataframe(df.reset_index(drop=True)).geometry[0])
'POLYGON ((-3.4181738244621784 50.63770319104376, -3.4183108033671856 50.637673939422605, -3.4183308056056596 50.63757996727042, -3.4182138292916653 50.637515246841126, -3.418076850806231 50.63754449846085, -3.4180568482152305 50.63763847051129, -3.4181738244621784 50.63770319104376))'