DWARF2 debug support patches + default i8086 arch (binutils side)

[asiekierka]

This merge request hopes to achieve two things:

• Add and use a default "i8086" architecture - this removes the need to specify -m i8086 in certain objdump calls, and also is necessary to effectively achieve the next point,
• Fix DWARF2 data alignment for IA-16.

This is a draft pull request, as while DWARF2 debug data is emitted, at least with regards to types and line numbers (with the proper GCC patch - see matching gcc-ia16 PR), and it looks correct to my eye, something still appears broken. (However, I do not think this is a blocker for potentially merging this PR, as no regressions are caused to my eye - while the GCC-side PR is dependent on the binutils-side PR, the opposite is not the case.)

[asiekierka]

I'm un-drafting this PR, as I have triaged the issue but need help figuring out a fix.

The issue which remains is with R_386_SUB32 relocations in segelf mode corrupting the debug sections' addresses. GCC primarily emits 32-bit relocations in i8086 builds as part of DWARF debug data; however, these pointers must point to the "linear" memory address of the symbol, and not the offset of the symbol. This is an issue in tc_gen_reloc, most likely, but I'm not familiar enough with binutils internals to have an idea of how to best fix this.

[asiekierka]

Also, if you'd prefer to receive merge requests on GitLab or Codeberg instead, please let me know and I'll do my best to move my upstreaming efforts there going forward.