I am applying mosaic's `grid_boundary` method on a spark DataFrame containing a set of `h3_hex_ids`. The geometries returned are not consistent. i.e they could be either `lat, long` or `long, lat`.
Here's a sample data
```
import pyspark.sql.functions as F
import h3
# Create a Spark DataFrame with a h3_hex_id column
df = spark.createDataFrame([
("point1", 612232698081050623),
("point2", 611987238200279039),
("point3", 612252103481491455)),
], ["name", "h3_hex_id"])
# Get Polygon geometries
df1 = df.withColumn("geometry", grid_boundary(col("h3_hex_id"), format_name=F.lit("WKT")))
df1.show()
+--------------------+------+----------------+----------------+
| name | h3_hex_id | | geometry |
+--------------------+------+----------------+----------------+
| point1 | 612232698081050623 | POLYGON ((42.1261...
| point2 | 611987238200279039 | POLYGON ((-145.83...
| point3 | 612252103481491455 | POLYGON ((-138.08...
+--------------------+------+----------------+----------------+
```
This is causing a few issues with downstream tasks like spatial joins. Why are the geometries returned by `grid_boundary` method not consistent?