Chore: BufferTransformer and Blowfish Refactor

[reneme]

This picks up the work started in #3942 by proposing a BufferTransformer class; which is essentially an orchestrator for BufferSlicer and BufferStuffer. As mentioned yesterday, I added some template fanciness, to handle the "block processing" necessary for various block ciphers. As suggested previously, this uses Blowfish to demonstrate the functionality (+ spanifying the blowfish implementation).

@randombit Is the process_blocks_of<> implementation in line with what you had in mind previously:

I wonder if we can fit in some other common functionality there, namely the word level loads/stores. That's a very common task (encrypt+decrypt for N ciphers plus alternative implementations)

Here's the idea of BufferTransformer::process_blocks_of, where the multi-block loop handling and buffer management is entirely abstracted. Also, as an additional goodie, we now have explicit compile-time knowledge of the blocked data length inside the processing lambdas. Whether the compiler uses this or not has to be seen. FWIW: I don't see a difference in the performance of Blowfish on my Mac.

void encrypt_n(const uint8_t inb[], uint8_t outb[], size_t blocks) {
   std::span<const uint8_t> in(inb, inb + blocks*BS);
   std::span<uint8_t> out(outb, outb + blocks*BS);
   
   BufferTransformer bt(in, out);
   bt.process_blocks_of<BS*4, BS>(
      overloaded{
         [](std::span<const uint8_t, BS*4> in, std::span<uint8_t, BS*4> out) {
            // encrypt four blocks at a time
         },
         [](std::span<const uint8_t, BS> in, std::span<uint8_t, BS> out) {
            // encrypt a single block
         }
      }
   );
}

No hard feelings if that doesn't make it into 3.5.0 at all. I just nerd-sniped myself yesterday and wanted to sketch this out. And frankly, I think the BufferTransformer implementation is quite concise. 😄

[coveralls]

coverage: 91.742% (+0.02%) from 91.72%
when pulling 7b3c762 on Rohde-Schwarz:rene/buffer_transformer
into 66216e1 on randombit:master.