feat: improve partitioning of credible sets

[d0choa]

The current credible sets after validation are completely shuffled. This can have a performance cost in all downstream processes.

In [7]: cs.select("chromosome", "position").show()
+----------+---------+
|chromosome| position|
+----------+---------+
|         2|151525381|
|         1|162388605|
|        10|102449815|
|         2| 47169327|
|        21| 42298501|
|         8| 85157259|
|         1| 11838976|
|        20| 45919050|
|         2|106566631|
|        16| 14889311|
|         1| 17005181|
|         2| 99176335|
|         1|149185498|
|         X|  1661217|
|         X|   312722|
|        11|   807149|
|        10| 79836221|
|         2| 73063040|
|        14| 55647604|
|         5| 43001499|
+----------+---------+
only showing top 20 rows

To improve the partitioning, this PR implements partitioning by range using chromosome and position.

The benefits on performance still need to be evaluated