Build abstraction to allow users to parse dumps directly from the file

[0vercl0k]

Build an abstraction that let the user be able to read the dump file the way they want:

• one reads from a memory mapped file (easier, maybe better perf?),
• one reads from the file on disk (better handling of very large file, lower memory footprint)

cc @neitsa for testing it once it's ready for a test-drive

[0vercl0k]

@neitsa I hacked up a small prototype in https://github.com/0vercl0k/kdmp-parser/tree/fbl_fileread - do you mind giving it a shot and reporting what you're seeing?

Running the tests, I am not seeing any real differences (at least on dump files <= 1gb) - but your experiments might be different.

Basically, I rewrote the parsing logic to call out to a Reader_t class which can be reading directly from a memory mapping or a file on the disk (FileReader_t).

Also cc'ing @hugsy as he might be interested.

Cheers

[0vercl0k]

Closing this as it's been stale