[DataFrame] Fully implement read_csv

[pschafhalter]

Changes

• Updates read_csv API to match Pandas 0.23
• Refactors code related to read_csv
• More efficient reading of CSV file
• More detailed warnings when defaulting to Pandas
• Bugfixes

Notes

• This PR depends on [DataFrame] Update initializations of IndexMetadata which use outdated APIs #2103 and [DataFrame] Fix bug with consistency between IndexMetadata and partitions #2088
• This PR assumes Pandas 0.23, so it will error until the rest of ray.dataframes is updated to support Pandas 0.23
• Resolves Deprecated parameters in read_csv method for pandas 0.23 #2082

Performance benchmarks

Performed on a 144MB CSV file.

Time to read

Current master

In [3]: %time df = rdf.read_csv("creditcard.csv")
CPU times: user 27.5 ms, sys: 5.23 ms, total: 32.7 ms
Wall time: 76.7 ms

PR:

%time df = rdf.read_csv("creditcard.csv")
CPU times: user 32.7 ms, sys: 12.3 ms, total: 45 ms
Wall time: 63.5 ms

Time to read and display

Pandas 0.23:

In [3]: %time df = pd.read_csv("creditcard.csv")
CPU times: user 3.1 s, sys: 123 ms, total: 3.22 s
Wall time: 3.22 s

Current master (Pandas 0.22):

In [2]: %%time
   ...: df = rdf.read_csv("creditcard.csv")
   ...: result = repr(df)
   ...: 
CPU times: user 160 ms, sys: 84.2 ms, total: 245 ms
Wall time: 2.63 s

PR:

In [5]: %%time
   ...: df = rdf.read_csv("creditcard.csv")
   ...: result = repr(df)
   ...: 
CPU times: user 153 ms, sys: 85.9 ms, total: 239 ms
Wall time: 2.18 s

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/5514/
Test PASSed.

[AmplabJenkins]

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/5607/
Test FAILed.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/5608/
Test PASSed.

[pschafhalter]

Moved to modin-project/modin#4.