small changes to get ready for pdf

[blublinsky]

Why are these changes needed?

There are currently 2 main issues for implementing pdf conversion:

1. PDF conversion is based on 2 files - *.csv and *.pdf. So we do need to process them as a pair. Both files have the same name but two different extensions. The solution is to only read *.csv file and then convert the name to get a pdf extension. Something like this:

pdf_name = TransformUtils.get_file_extension(file_name)[0] + ".pdf" 

and then read a pdf file

2. As PDF is doing conversion from *.csv to *.parquet, the checkpointing is broken. Here checkpointing is fixed by:
   a. introducing a new parameter files_to_checkpoint. default is [.parquet]
   b. comparing just the full names without extensions for the purposes of checkpointing

Related issue number (if any).

[dolfim-ibm]

Thanks @blublinsky