Inconsistent `codeOffset` parsing

[cpebble]

I encountered an inconsistency when working with E-T. When parsing negative code offsets, conditional jumps will happily accept this, however normal jumps throw a parser error.

Example:

; Valid
jgt r0, r11, +-5
; Invalid
ja +-5

This also leads to an issue in Ebpf.Display where displayProgram generates code it cannot parse:

>>> displayProgram [Jmp (-3)]
"ja +-3\n"
>>> parse "ja +-3\n"
Left "\"<input>\" (line 1, column 5):\nunexpected \"-\"\nexpecting \"0\" or digit"

You seem to have noticed this here: AsmParser.hs:139

IMO the easiest patch would be to include the required '+' in jmp parsing. This syntax is how Ken presented it in the Software Security course.

[kfl]

Thanks for the report. You are completely right, it is inconsistent and should be fixed.