Stdlib: Reorganize build, test and nix setup

[nomeata]

Noteworthy changes:

• Stdlib sources are in stdlib/src
  tests are in stdlib/test
  docs are in stdlib/doc
  (no more rm $out/*Test.mo)

• Individual Makefiles in these directories,
  as well as the three examples in stdlib/examples/*

• In stdlib/tests, auto-generate a test that simply imports
  all stdlib files. This makes sure the stdlib actually compiles
  (was not the case before)

• Tests and examples use package imports "mo:stdlib/…" to import
  the stdlib, instead of relative paths.
  The Makefiles are set up to point to ./stdlib/src by default. But
  when building via nix, we are building against the stdlib as
  installed by nix

• More fine-grained nix setup, so when you just change the examples,
  or the tests, the stdlib derivation does not have to be rebuild.

• Runs the tests for Rx and ActorSpec using wasmtimes

[nomeata]

(if someone is worried about possible merge conflicts of their PRs against stdlib, I volunteer to merge master into them after this lands.)

[matthewhammer]

This is great; thanks, Joachim!

[matthewhammer]

LGTM

[ggreif]

Let me try these.

[ggreif]

Unfortunately the .md, .html files are garbled?

import Prim "mo:prim";
import Iter "iter.mo";

module {
Hash values

Try

head ~/motoko/stdlib/doc/_out/hash.md

Similarly

tail ~/motoko/stdlib/doc/_out/hash.md

@matthewhammer maybe you can adapt the doc extractor to deal with the new-fangled imports and module { ... } wrapping? (Could be in a follow-up PR!)

[nomeata]

Is that a regression with the docs?

I am tempted to work on a simple doc system that is good enough to get the stdlib docs into the current manual soon (even if it's not anything we can hand out to people to use on their own libraries yet). So I'd rather not hold this up by the (currently unused) docs.

[ggreif]

Yah, polished docs can wait. LGTM!

[nomeata]

Closed in favor of #1135

[matthewhammer]

@matthewhammer maybe you can adapt the doc extractor to deal with the new-fangled imports and module { ... } wrapping? (Could be in a follow-up PR!)

I like the idea of fixing this (it should be fixed), but I don't like the idea of investing more energy in my python script. It was never meant to live this long, honestly.

What did we decide this morning, in the team meeting?

[nomeata]

We didn't decide anything really. I am inclined to have a closer look at what we have, maybe refine the hack we have, and produce asciidoc that can be included as a chapter in the manual maybe.