Add to_samples and from_samples

[penelopeysm]

See TuringLang/DynamicPPL.jl#1087 (comment), this is the implementation of option (1) from that comment, as discussed in our developer meeting earlier.

I think it makes more sense to use to_samples and from_samples instead of from_chains and to_chains, it reads better.

There are additional bits of AbstractChains interface in #180, but I don't want that to hold this up, because this functionality is semi-urgently needed in DynamicPPL.

Specifically, there is currently no easy way to obtain a Dict of parameters from MCMCChains. Combined with the fact that you can now only reliably initialise parameters using InitFromParams(::Dict), this means that there is no easy way to initialise sampling from one row of MCMCChains.Chains.

[github-actions]

AbstractMCMC.jl documentation for PR #184 is available at:
https://TuringLang.github.io/AbstractMCMC.jl/previews/PR184/

[mhauru]

I agree on this being a better name.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 0.00%. Comparing base (e58f735) to head (12f8bf0).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@     Coverage Diff     @@
##   main   #184   +/-   ##
===========================
===========================

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.