add support for datasketches js udafs using wasm libraries

[nikunjbhartia]

Adding support for 3 datasketches sketch types : 1) Theta sketch 2) KLL sketch 3) Tuple sketch

Total functions:

• 9 Aggregate functions ( JS UDAFs)
• 7 scalar functions ( JS UDF)

Solution approach: Datasketch cpp libraries are compiled to WASM libraries using emscripten toolchain and loaded in BQ JS UDAFs. using docker.io emscripten image in build file

Note: currently unit tests are executed using dataform test which do no support UDAFs or constant arguments. Hence, added test cases for 7 scalar functions only

There is also a datasketches-cpp directory which could be added as a submodule, committed all the files for now, lets discuss during review.

[danieldeleo]

I've submitted https://github.com/GoogleCloudPlatform/bigquery-utils/pull/425 to enable unit testing for UDAFs. Once it's merged, please pull the changes and add the remaining 9 test cases for the UDAFs

[nikunjbhartia]

Thanks for adding support for UDAF unit testing. Current framework uses positional args which would fail for functions having const args. we need to add additional support for passing constant args which are defined as "NOT AGGREGATE" in UDAF function signature.

I have added unit test for 1 UDAF which did not have any const arg in its signature. Rest of the 8 UDAFs have const args and hence relevant tests can't be added yet.

[danieldeleo]

@nikunjbhartia you should be good to go for the rest of the 8 UDAFs now that https://github.com/GoogleCloudPlatform/bigquery-utils/pull/426 is merged

[nikunjbhartia]

Thanks @danieldeleo for the quick fix. Made following updates:

• Added unit tests for rest of the 8 UDAFs with const args
• Removed datasketches-cpp folder
• Added a build task for cloning the above repo

[danieldeleo]

/gcbrun