Include portable and non-portable assembly files

[jtraglia]

First and foremost, sorry about how big this PR is. Most of the changes are in generated files though. I'm open to any alterations you might have in mind. I want to make these changes in preparation for CPU runtime detection; if not in this repository, then to minimize the differences with a fork that does.

The biggest part of this PR adds a prefix (${pre}) to the beginning of labels for internal routines for assembly files that have multiple implementations. This allows us to include both portable & non-portable implementations. Then select the right implementation based on the defines.

As an overview, this will:

• Add prefixes to internal functions in these files:
  • ct*_inverse_mod_384-x86_64.pl
  • mul*_mont_256-x86_64.pl
  • mul*_mont_384-x86_64.pl
• Remove #if defined(...) checks in assembly.S.
  • So both implementations will be available.
• Remove sha256_emit, sha256_bcopy, sha256_hcopy from non-portable implementation.
  • These routines are the exact same in sha256-x86_64.pl and sha256-portable-x86_64.pl.
• Rename K256 table in portable implementation, it conflicted with the other one.
• Rename sha256_block_data_order in the portable file to sha256_block_data_order_portable.
  • This conflicted with the label in the other assembly file.
• Include mach-o/sha256-portable-x86_64.s for apple/x86_64.
  • Guess this wasn't necessary before?
• Build/include sha256-portable-x86_64.s for win64.
  • Guess this wasn't necessary before either?
• Alter sha256_block "selector" preprocessor code.
  • Needed to differentiate between portable/non-portable for x86.
  • Thought this was a little more clear.

Other considerations:

• The size of the built library (libblst.a) has increased from 243K to 268K, or ~10% bigger.
  • I'm actually surprised it's not more of a difference, shows how small the assembly is.

Also, for most of the replacements I did this in vim, for each of those perlasm files:

:%s/___/FOOBARS/g
:%s/__/${pre}__/g
:%s/FOOBARS/___/g

[dot-asm]

I'm inclined to say that you're overthinking it. I'd rather do it as in #165. The referred PR is just the very first step, more steps to come...

[jtraglia]

Yeah that PR looks much simpler. Let's go that route 👍