Add PSyclone Transmute Step

The current psyclone step is pretty much hardcoded to only support DSL, not a transmute step (besides pre-processing X90 files, it also parses kernel files to get the checksum of the metadata, handles two output filenames, while a transmute step has only one step. It handles kernel search path, which is not required for transmute).

While the existing code could be augmented with additional if-statements to support both use cases, it would appear much cleaner if they are just separate steps (maybe with a common base class??)

Start by examining the existing PSyclone step and the processing described in the issue, then compare those requirements with the transmute step's single-step flow. The work is done when transmute has a separate step that avoids DSL-only parsing, kernel search paths, and multiple output filenames while preserving the existing PSyclone behavior.

Written by the indexing model from the issue text.