[FEA] Tool output file support for writing to HDFS file system

[wjxiz1992]

Is your feature request related to a problem? Please describe.
This is from a customer request.

Background:
The customer is trying to qualify all their current Spark jobs. The number of spark jobs are so many that they have to write a customized UDF to make the qualification work distributed: #1347

In the PR above, the general IDEA is to do a two step thing in the UDF:

1. Leverage Scala qualification code to get qualification output.
2. Call qualx command line to do predictions with the output files produced in step-1

Step-1 has been proved not valid, as some essential files for step-2 doesn't produced by Scala code. It requires also the python side code. Also it doesn't support hdfs as output filesystem

spark_rapids_tools.exceptions.InvalidProtocolPrefixError: "hdfs:///..../..../...." is not a valid path since it does not start with "file://"

Step-2 only accepts local file system, which is inconvinient for distributed work in their production environment.

Thus, this issue is created to ask for full-chain support for HDFS file system. The command line after this feature is supported can be imagined like below:

spark_rapids qualification \
--platform onprem \
--eventlogs hdfs:///<PATH_TO_EVENTLOG_FILE> \
--output_folder hdfs:///<PATH_TO_OUTPUT_PATH>


spark_rapids prediction \
--platform onprem \
--qual_output hdfs:///<PATH_TO_OUTPUT_PATH_PRODUCED IN QUALIFICATION>\
--output_folder hdfs:///<PATH_TO_OUTPUT_PREDICTIONS>

This should contribute to #1304

cc @Heatao @winningsix