Use Python type name instead of Spark's in error messages.

[ueshin]

Addressing #1980 (comment) to add pandas dtypes.

[codecov-io]

Codecov Report

Merging #1985 (c8520db) into master (5f27857) will not change coverage.
The diff coverage is 45.45%.

@@           Coverage Diff           @@
##           master    #1985   +/-   ##
=======================================
  Coverage   94.55%   94.55%           
=======================================
  Files          50       50           
  Lines       10939    10939           
=======================================
  Hits        10343    10343           
  Misses        596      596           

Impacted Files	Coverage Δ
databricks/koalas/generic.py	92.61% <33.33%> (ø)
databricks/koalas/frame.py	96.73% <100.00%> (ø)
databricks/koalas/series.py	96.75% <100.00%> (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 5f27857...c8520db. Read the comment docs.