Avoid overwriting a variable with value of different type.

[obi1kenobi]

Description

Avoids the following mypy error:

arviz/plots/traceplot.py:183: error: "InferenceData" has no attribute "dims"  [attr-defined]

Part of #1496.

The error was happening because the data variable was initially of type InferenceData but was later overwritten with a value of different type on the line:

data = get_coords(convert_to_dataset(data, group="posterior"), coords)

In general, assigning a value of a completely different type to an existing variable will cause both mypy and pylint to complain about the assignment itself. In this case, the get_coords() function itself was not type-hinted, i.e. its return type was considered to be Any. This led mypy to incorrectly deduce that the return type must have preserved the existing type of data, because not doing so would have been an immediate error.

Renaming the assignment on that line to a new variable name avoids the current error as well as the possible future mypy or pylint errors arising when and if get_coords() gets a type hint for its return type.

Checklist

• Follows official PR format
• Code style correct (follows pylint and black guidelines)

[codecov]

Codecov Report

Merging #1505 (cc96b1f) into master (14875f5) will not change coverage.
The diff coverage is 80.00%.

@@           Coverage Diff           @@
##           master    #1505   +/-   ##
=======================================
  Coverage   91.76%   91.76%           
=======================================
  Files         105      105           
  Lines       11316    11316           
=======================================
  Hits        10384    10384           
  Misses        932      932           

Impacted Files	Coverage Δ
arviz/plots/traceplot.py	95.12% <80.00%> (ø)

---

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 14875f5...cc96b1f. Read the comment docs.

[ColCarroll]

LGTM -- this is interesting, because InferenceData is not an xarray dataset, but it pretends to be. I guess this is actually a helpful distinction, because data has properties hanging off it, while data.posterior does not.