squeeze dim_0 dimensions for scalars away

[TimOliverMaier]

1. Squeeze scalars in xarray.Dataset generated for constant_data by arviz. This is a provisional fix until the behavior of arviz here is refactored.

What is this PR about?
This PR is supposed to fix #6755 , by squeezing the _dim_0 dimension in the constant_data dataset for all scalars.
Checklist

• Explain important implementation details 👆
• Make sure that the pre-commit linting/style checks pass.
• Link relevant issues (preferably in nice commit messages)
• Are the changes covered by tests and docstrings?
• Fill out the short summary sections 👇

Bugfixes

• scalars are now stored dimensionless in constant_data. Fix of Scalar constant data is promoted to rank 1 in returned InferenceData #6755

---

📚 Documentation preview 📚: https://pymc--6764.org.readthedocs.build/en/6764/

[codecov]

Codecov Report

Merging #6764 (654e9f9) into main (261862d) will decrease coverage by 0.02%.
The diff coverage is 90.24%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6764      +/-   ##
==========================================
- Coverage   91.93%   91.92%   -0.02%     
==========================================
  Files          95       95              
  Lines       16125    16139      +14     
==========================================
+ Hits        14825    14835      +10     
- Misses       1300     1304       +4     

Impacted Files	Coverage Δ
pymc/logprob/basic.py	96.85% <55.55%> (-3.15%)	⬇️
pymc/backends/arviz.py	96.48% <100.00%> (+0.09%)	⬆️
pymc/logprob/__init__.py	100.00% <100.00%> (ø)
pymc/logprob/scan.py	98.49% <100.00%> (ø)
pymc/model.py	89.91% <100.00%> (ø)
pymc/sampling/mcmc.py	89.88% <100.00%> (+0.06%)	⬆️
pymc/step_methods/compound.py	97.63% <100.00%> (ø)
pymc/testing.py	91.60% <100.00%> (ø)

[ricardoV94]

Thanks @TimOliverMaier