Re-export certain functionality from DynamicPPL

[torfjelde]

Given that Turing.jl has become a "public-facing" shell package, I think it's worth exporting some functionality commonly used with models from DynamicPPL.jl, so the end-user doesn't have to see DynamicPPL to be able to use these.

Amongst these I think the following should be included:

• condition
• decondition
• fix
• unfix
• OrderedDict (the only reliable way to call many methods, e.g. rand)
• conditioned / observations (though probably just one of these?)

Any objections? @yebai @devmotion

[yebai]

Looks good to me -- let's export conditioned instead of observations for consistency with condition/decondition APIs

[coveralls]

Pull Request Test Coverage Report for Build 9141352740

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 0.0%

---

Totals
Change from base Build 9140976557:	0.0%
Covered Lines:	0
Relevant Lines:	1529

---

💛 - Coveralls

[torfjelde]

Okay, these tests failing are now becoming somewhat worrying (previously everything passed except on one run, but now it's also failing on other systems)

[torfjelde]

Why was this merged?

[torfjelde]

@yebai

[torfjelde]

I specifically said "please don't do this" just a few days ago.

[torfjelde]

I see tests are at least passing after the merge, but still, please don't do this. Ping the creator of the PR to check first.

[yebai]

Let's be flexible about merging. Core members should be allowed to merge PRs considered "ready to go" -- it's debatable what can be treated as "ready to go", but this PR probably fits into that group safely I think.

[torfjelde]

But what arguments exist for merging vs. pinging the creator of the PR and asking them to merge @yebai ? I can't see any in favour of doing this.

In contrast, I can easily come up with arguments for not doing this, e.g. we've had examples before where the PR creator realizes the test cases aren't sufficient, starts working on it, tries to push, only to realize the branch has already been merged and released with bugs the PR creator was in the process of fixing.

[torfjelde]

But regardless of whether we want to make "don't merge someone else's PR without checking with them first" a team-wide rule or not, I ask that you at least don't merge PRs that I have a created but instead ping me and ask if it's ready to merge.
If I'm being very unresponsive, then sure, feel free to merge, but otherwise let me have the final say in whether a PR I have made is ready or not.