Converting pandas index takes very long, add in arrow_table. (#41)

* Converting pandas index takes very long, add in arrow_table.

* Bump pyproject toml version number for pip build.

Author: justinGilmer

btrdb/transformers.py

@@ -224,10 +224,7 @@ def arrow_to_dataframe(
         tmp = tmp_table.select(["time", *usable_cols])
     else:
         tmp = tmp_table
-    df = tmp.to_pandas(date_as_object=False, types_mapper=pd.ArrowDtype)
-    df = df.set_index("time")
-    df.index = pd.DatetimeIndex(df.index, tz="UTC")
-    return df
+    return tmp.to_pandas(date_as_object=False, types_mapper=pd.ArrowDtype)
 
 
 def to_dataframe(streamset, columns=None, agg="mean", name_callable=None):
@@ -668,5 +665,7 @@ class StreamSetTransformer(object):
     to_polars = to_polars
     arrow_to_polars = arrow_to_polars
 
+    arrow_to_arrow_table = arrow_to_arrow_table
+
     to_csv = to_csv
     to_table = to_table

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "btrdb"
-version = "5.30.1"
+version = "5.30.2"
 authors = [
     {name="PingThingsIO", email="support@pingthings.io"},
 ]