[BUG] GpuRegExExtract is not align with RegExExtract

[sperlingxx]

Describe the bug

1. In Spark, the pattern of regexp_extract isn't required to fully match the input string. Just as what extract means, regexp_extract finds any substring which matches the pattern and extracts them. However, GpuRegExExtract requires a full match on pattern.

val df =Seq("1a", "2a", "3a", "4a", "5a", "6a", "7a", "8a", "9a", "10a").toDF("c")
df.coalesce(1).select(regexp_extract(col("c"), "(a)", 1)).collect()

GPU result: Array([], [], [], [], [], [], [], [], [], [])
CPU result: Array([a], [a], [a], [a], [a], [a], [a], [a], [a], [a])

2. When group index is 0, GpuRegExExtract doesn't behave correctly.

CPU run: regexp_extract('123abcEfg', '([0-9]+)[a-z]+([A-Z])', 0) => 123abcE
CPU run: regexp_extract('123abcEfg', '([0-9]+)[a-z]+([A-Z])', 0) => 123abcEfg

BTW, the issue is originated from #5088.