Merge work about sections and pointer offsets

[Mesabloo]

Implements working basic sections (code and data, but this can easily be extended to others) as described in #19, #40 and #2.

Summary of the PR:

• [x] implement code and data sections:
  • [x] parse each section individually with different grammars
  • [x] post-process the AST to concatenate all same sections
  • [x] import all symbols in data sections in the typechecker context
  • [x] typecheck instructions only in a code section
  • [x] compile the data section of the typed AST along with the code section
• [x] implement pointer offsets
  • [x] parse pointer offsets
  • [x] typecheck pointer offsets according to the specification
  • [x] post-process (after type-checking) the typed AST to desugar complicated pointer offsets (such as %r0(%r1))
  • [x] compile the desugared instructions
  • [x] push @r64
  • [x] pop @r64
  • [x] add @r64 @r64
  • [x] mov @*a @r64
  • [x] add @s64, @r64
  • [x] mov @(r64-N) @r64
  • [x] sub @s64 @r64
  • [x] compile relocation information for data label offsets

This PR also adds marginal functionalities such as:

• [x] dumping the untyped and typed ASTs to different files using the two -ddump-ast and -ddump-typed-ast options
• [x] allow choosing the output object file name with the -o FILENAME command-line option
• [x] rewriting most of the ELF file generation library in C (and generate a Doxyfile for the documentation)

Once all of this is done, the PR can be merged.