bug: Support read_parquet glob file paths

[petern48]

This one was pretty annoying to debug. Here's what was going on. We were calling .to_urls() in twice: once in geoparquet_listing_table and once in here in context.rs's read_parquet:

sedona-db/rust/sedona/src/context.rs

Lines 229 to 235 in 5d3cfba

	pub async fn read_parquet<P: DataFilePaths>(
	&self,
	table_paths: P,
	options: GeoParquetReadOptions<'_>,
	) -> Result<DataFrame> {
	let urls = table_paths.to_urls()?;
	let provider =

The second call was causing the ListingTableUrl's glob attribute to being removed. I fixed this my removing one of the calls.

Note: globbing with the * operations still does NOT work on AWS s3 paths. That would need an additional fix / feature. It looks like Datafusion doesn't support globbing s3 paths yet either: It looks like DataFusion doesn't support globbing s3 paths yet either: apache/datafusion#16303

[paleolimbot]

Remind me that I owe you cookies or chocolate cake someday for debugging this one.

Thank you!