ARROW-2626: [Python] Add column name to exception message when writing pandas df fails

[louispotok]

Goal is to make debugging easier - it's now possible to see which column failed.

One quick question, I'm not sure if we need to be catching both ArrowInvalid and ArrowTypeError - I think this used to throw ArrowInvalid but now only throws ArrowTypeError. Thoughts?

[codecov-io]

Codecov Report

Merging #2075 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2075      +/-   ##
==========================================
- Coverage   86.33%   86.31%   -0.02%     
==========================================
  Files         242      230      -12     
  Lines       41101    40217     -884     
==========================================
- Hits        35485    34714     -771     
+ Misses       5616     5503     -113

Impacted Files	Coverage Δ
python/pyarrow/tests/test_convert_pandas.py	94.8% <100%> (+0.01%)	⬆️
python/pyarrow/pandas_compat.py	97.85% <100%> (+0.02%)	⬆️
rust/src/list.rs
rust/src/record_batch.rs
rust/src/builder.rs
rust/src/error.rs
rust/src/datatypes.rs
rust/src/memory.rs
rust/src/bitmap.rs
rust/src/array.rs
... and 4 more

---

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 fcc13f5...cf23778. Read the comment docs.

[wesm]

+1, thank you. From looking at python/numpy_to_arrow.cc, we are returning both TypeError and Invalid for various failure modes. I will open a JIRA about looking through these errors to decide if we are returning the right error types

[wesm]

See https://issues.apache.org/jira/browse/ARROW-2702