add MultiObservedRVs to observed_data

[OriolAbril]

Description

closes #384. Data from MultiObservedRV will now be added to observed_data group by from_pymc3

Checklist

• Follows official PR format
• New features are properly documented (with an example if appropriate)?
• Includes new or updated tests to cover the new feature
• Code style correct (follows pylint and black guidelines)
• Changes are listed in changelog

[codecov]

Codecov Report

Merging #1098 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1098   +/-   ##
=======================================
  Coverage   93.14%   93.14%           
=======================================
  Files          94       94           
  Lines        9497     9500    +3     
=======================================
+ Hits         8846     8849    +3     
  Misses        651      651           

Impacted Files	Coverage Δ
arviz/data/io_pymc3.py	93.33% <100.00%> (+0.07%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ec33b4c...17f15b7. Read the comment docs.

[OriolAbril]

I have updated code and tests to get from_pymc3 to store data from MultiObservedRV in observed_data group

[canyon289]

@OriolAbril Feeling good about this one? Happy to review and merge if so

[OriolAbril]

Yep, should be ready to merge now.

[canyon289]

Merging

[ClaireGyn]

I have updated code and tests to get from_pymc3 to store data from MultiObservedRV in observed_data group

Hi, OriolAbril. Does that work now? I just tried to implement your code according to the way as shown in the screenshots. But it seems it reports the same error. The MultiObservedRV object can't be transferred normal values which can be presented in the graph.

[OriolAbril]

It should work with the latest ArviZ versions, and IIRC we have some tests to check this works so I have no reason to suspect a regression here for now, there are many other possible issues with dtypes and data conversions.

What is the code that generates the model? I would recommend asking a question on PyMC3 discourse (given that you are using pymc, otherwise Stan discourse for example would be equally fine) including the code and the error message so it's easier to find in case someone has the same issue again. Feel free to tag me there, I have the same handle. If you manage to reduce the issue to a minimal reproducible example we can use it to play around and see what the proposed solutions do.