Lots of error cases are untested, which is worth improving but not urgently so.
Here are some chunks of code that should better be tested (line numbers as found in rev 0448bfbb9a1845a3cca5d07a32fdd4590f538713, [HTML coverage report](https://github.com/cython/cython/suites/2745137547/artifa

Problem: the approximate method can still be slow for many trees
catboost version: master
Operating System: ubuntu 18.04
CPU: i9
GPU: RTX2080

Would be good to be able to specify how many trees to use for shapley. The model.predict and prediction_type versions allow this. lgbm/xgb allow this.

change all x-moloch-cookies to x-arkime-cookies in tests and middleware 😋

There is no technical difficulty to support includeValue option, looks like we are just missing it on the API level.

See SO question

... to make it easier to read Vespa documentation on an e-reader / offline

Vespa documentation is generated using Jekyll from .md and .html files, look into options for generating the artifact as part of site generation (there might be plugins we can use here)

Currently, there is no authentication in our MongoDB tests.
Setting MONGO_INITDB_ROOT_USERNAME and MONGO_INITDB_ROOT_PASSWORD environment variables enable auth in MongoDB docker. https://hub.docker.com/_/mongo

Hi, if my spark app is using 2 storage type, both S3 and Azure Data Lake Store Gen2, could I put spark.delta.logStore.class=org.apache.spark.sql.delta.storage.AzureLogStore, org.apache.spark.sql.delta.storage.S3SingleDriverLogStore

Thanks in advance

Use case:
Right now one can only use date_trunc() to easily define time buckets. date_trunc() only supports predefine time intervals like 1 minute, 1 hour, etc. . In time-series use cases it is often necessary to define different time bucket sizes like e.g. '5 minutes' or '20 minutes'

a workaround for this is the - error prone - integer division on the timestamp e.g.

S