Add Libseff Benchmarks

[codrutiftode]

I have implemented the following benchmarks in libseff:

• countdown
• fibonacci recursive
• product early
• iterator
• generator
• parsing dollars
• resume nontail
• handler sieve

I have also added a Dockerfile and updated the root level Makefile as to allow for building and running a libseff container. Have a look and let me know what you think! I tried to maintain a balance between staying close to the other implementations and making use of idiomatic C structures where possible (loops instead of recursion where the required concept was just a simple loop).

Finally, I updated the README as required. I placed a minus sign for nqueens, tree explore and triples, because they require multi-shot continuations which libseff does not support as far as I know.

Note: the libseff repo (https://github.com/effect-handlers/libseff.git) makes use of clang-10, so I have mirrored this in the Dockerfile / Makefiles, although the associated paper references clang-12. This required using Ubuntu 18.04 instead of ubuntu-latest as the base Docker image.

[dhil]

Looks good. I haven't studied it in great detail yet. A few things stand out:

• You seem to always count backwards in loops, e.g. i = N; i > 0; i--. I suggest counting forward instead, it matters for performance when you use the iterator variable i to index into an array.
• Mark non-public functions with static, i.e. every function except for main, I think.

[jasigal]

Thanks everyone for your input, I've looked everything over one last time and it LGTM :)