Protobuf as bytearray for spark conf

[mlathara]

We were passing in protobuf as json string in for spark. It turns out that Google's python protobuf sends in int64 as string (for reasons specified in many places, for instance here)

So this moves us to expect protobuf bytestream instead. Since we use hadoop configuration to pass the config, it gets encoded as a base64 string on the sender side - so the receiver decodes that first before ingesting the protobuf object.

Also, a few annoying travis Mac OS changes.

[codecov]

Codecov Report

Merging #76 into develop will decrease coverage by 28.36%.
The diff coverage is 63.63%.

@@              Coverage Diff               @@
##             develop      #76       +/-   ##
==============================================
- Coverage      77.49%   49.13%   -28.37%     
- Complexity         0      412      +412     
==============================================
  Files            117      111        -6     
  Lines          16062    14974     -1088     
  Branches         330      330               
==============================================
- Hits           12447     7357     -5090     
- Misses          3426     7431     +4005     
+ Partials         189      186        -3

Impacted Files	Coverage Δ	Complexity Δ
...a/org/genomicsdb/reader/GenomicsDBQueryStream.java	48.71% <ø> (ø)	10 <0> (+10)	⬆️
...g/genomicsdb/reader/GenomicsDBFeatureIterator.java	79.26% <ø> (ø)	17 <0> (+17)	⬆️
...csdb/importer/extensions/CallSetMapExtensions.java	40% <ø> (-21.34%)	8 <0> (+8)
...ain/java/org/genomicsdb/spark/GenomicsDBInput.java	62.31% <100%> (-4.98%)	41 <0> (+41)
.../org/genomicsdb/spark/GenomicsDBConfiguration.java	57.33% <50%> (-2.81%)	20 <0> (+20)
...va/org/genomicsdb/spark/GenomicsDBInputFormat.java	63.33% <60%> (ø)	8 <0> (+8)	⬆️
src/main/cpp/include/api/genomicsdb.h	0% <0%> (-100%)	0% <0%> (ø)
src/main/cpp/src/api/genomicsdb.cc	0.41% <0%> (-95.4%)	0% <0%> (ø)
...rc/main/cpp/src/genomicsdb/genomicsdb_iterators.cc	0.27% <0%> (-94.45%)	0% <0%> (ø)
...cpp/include/genomicsdb/genomicsdb_columnar_field.h	0% <0%> (-93.45%)	0% <0%> (ø)
... and 70 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 49e2773...e464055. Read the comment docs.

[nalinigans]

Looks good to me.

[kgururaj]

Looks good to me

• I'm assuming the reason to convert back and forth between a bytearray and String is that Spark configuration doesn't accept binary values.
• Facepalm for the string int64 representation in JSON to deal with Javascript

[mlathara]

Yup - Spark conf doesn't take binary values.