Change default write mode to overwrite to be consistent with pandas

[HyukjinKwon]

pandas always overwrites. Currently, only JSON and CSV overwrites in Koalas. This PR proposes to consistently overwrite for now.

>>> import pandas as pd
>>> pd.DataFrame({'a':[1,2,3]})
   a
0  1
1  2
2  3
>>> pd.DataFrame({'a':[1,2,3]}).to_parquet("/tmp/abc.parquet")
>>> pd.DataFrame({'a':[1,2,3]}).to_parquet("/tmp/abc.parquet")
>>> pd.DataFrame({'a':[1,2,3]}).to_parquet("/tmp/abc.parquet")

[codecov-io]

Codecov Report

Merging #1209 into master will decrease coverage by 1.36%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1209      +/-   ##
==========================================
- Coverage   95.18%   93.82%   -1.37%     
==========================================
  Files          35       35              
  Lines        7201     7201              
==========================================
- Hits         6854     6756      -98     
- Misses        347      445      +98

Impacted Files	Coverage Δ
databricks/koalas/frame.py	96.96% <100%> (ø)	⬆️
databricks/koalas/usage_logging/__init__.py	24.32% <0%> (-72.98%)	⬇️
databricks/koalas/usage_logging/usage_logger.py	50% <0%> (-50%)	⬇️
databricks/koalas/__init__.py	78.72% <0%> (-6.39%)	⬇️
databricks/conftest.py	94% <0%> (-4%)	⬇️

---

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 535424b...7062bb8. Read the comment docs.

[ueshin]

I noticed the other but related thing.

[ueshin]

oh,, should this lines be self.to_spark().write.mode(mode).saveAsTable(...)?

[ueshin]

And ditto for other to_xxx related to Spark IO?

[HyukjinKwon]

Eh, do you mean for consistency? saveAsTable in Spark seems having mode.

[HyukjinKwon]

I just made the changes back.

[ueshin]

nvm, I thought we should explicitly specify mode with write.mode(mode), but it's not needed.

[ueshin]

LGTM.

[ueshin]

Thanks! merging.