[FEA] Support GetMapValue with non-literal key

[viadea]

I wish we can support GetMapValue with non-literal key.

For example:

from pyspark.sql.types import *
from pyspark.sql.functions import *
jsonString="""{"Zipcode":704,"ZipCodeType":"STANDARD","City":"PARC PARQUE","State":"PR","Zipcode":999}"""
df=spark.createDataFrame([(1, jsonString)],["id","value"])
df2=df.withColumn("value",from_json(df.value,MapType(StringType(),StringType())))

df2.withColumn('newcol', expr(f""" map_filter(value, (k,v)  -> value[split(k, "abc")[0]] is not NULL    )       """) ).collect()

Not-supported messages:

!Expression <GetMapValue> value#136[split(lambda k#225, abc, 2)[0]] cannot run on GPU because key only supports StringType if it is a literal value

[sameerz]

@mythrocks is this a duplicate of #5204 ?

[mythrocks]

Yep, it's a dupe. The libcudf end of it supports only scalar keys, currently.

It's not hard to add support for vectors. Just needs doing.