Improve simulation UX

[survived]

Writing tests for MPC protocols involves copy-pasting the same boilerplate over and over again. This PR improves ergonomics by moving the most common logic into the lib.

You can see how it improves the tests in LFDT-Lockness/cggmp21#120, specifically in 81191c0:

• A lot less code required:
  • No need to manually poll futures via futures::future::try_join_all
  • No need to write a check each time that all outputs are equally the same (e.g. in signing, we need to check that all sigs are the same, and we used to write a loop each time, now we just use .expect_eq())
• No need of async runtime. All tests are sync now.

Other changes:

• Remove std feature. All errors implement core::error::Error trait.

[maurges]

One part of me wants to say that we should keep the old api, on the other hand I'm not going to use it anywhere.

One nitpick is that I'd prefer using "simulation" instead of "sim"

[survived]

One part of me wants to say that we should keep the old api

How do we improve UX and evolve the lib if we commit to keep old api? 😃

[survived]

One nitpick is that I'd prefer using "simulation" instead of "sim"

I also thought about it, though writing "simulation" wasn't an issue when I was updating all the dependent lib... I think I'll go ahead and use shorter name

[survived]

@maurges that's a bit out of scope of the PR, but I also removed uses of std::error::Error trait, and replaced with core::error::Error in 5d53357

[survived]

Too soon! @maurges Clippy suddenly started complaining on unmodified part of the library

[survived]

That's a large PR...