Add a build in CI for pyspark 3.0

[liangz1]

The pypi package is not released yet. Once it is available, I'll update the code.

[codecov[bot]]

Codecov Report

Merging #521 into master will increase coverage by 0.05%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #521      +/-   ##
==========================================
+ Coverage   86.18%   86.24%   +0.05%     
==========================================
  Files          81       81              
  Lines        4467     4471       +4     
  Branches      717      718       +1     
==========================================
+ Hits         3850     3856       +6     
+ Misses        505      503       -2     
  Partials      112      112              

Impacted Files	Coverage Δ
petastorm/unischema.py	95.79% <100.00%> (+1.03%)	:arrow_up:

---

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 4df17f3...2cab881. Read the comment docs.

[WeichenXu123]

@liangz1 I build latest master pyspark and upload to S3. Use pip install pip install https://ml-team-public-read.s3-us-west-2.amazonaws.com/pyspark-3.1.0.dev0-60dd1a690fed62b1d6442cdc8cf3f89ef4304d5a.tar.gz to install latest spark.

[WeichenXu123]

Summary:

• This PR add latest spark master version test in CI.
• Fix dict_to_spark_row method, See https://github.com/uber/petastorm/pull/521#discussion_r400610879
• For spark<3.0 and pyarrow>=0.15, add ARROW_PRE_0_15_IPC_FORMAT=1 in CI, and remove unnecessary skipif in test. See explanation in https://spark.apache.org/docs/3.0.0-preview/sql-pyspark-pandas-with-arrow.html#compatibiliy-setting-for-pyarrow--0150-and-spark-23x-24x