Fixes, tests, and enhancements for grep (part 1)

The grep implementation here needs some TLC.

In this MR:

• Use standard exit behavior:
  • Return 0 if there were matches and no errors (and for --help)
  • Return 1 if there were no matches (useful in scripts)
  • Return >1 if there were any errors, but wait to do so until all files have been processed (useful when globbing over a busy directory, for example)
• Fix a couple bugs:
  • Don't throw a missing pattern error when --help is called
  • Suppress match output if --count is given
• Add filename headers when dealing with multiple files (and standard -h/-H flags)
• Implement --ignore-case, --max-count, and --quiet flags
• Add fairly comprehensive integration/regression tests
• Switch to arg-parser library for argument parsing

There's more I'd like to do (regex support, recursive search, options for context and binary files), but I don't want this MR to get too big, especially since it's my first Redox contribution.