M1 support, upgrade to newer RocksDB and AWS Java SDK

[DevinTDHa]

Description

This PR enables Spark NLP to be built with M1 support. Most of the required packages have already been published and have been integrated into the build.

This PR fixes the remaining compatibility issues:

• Add another condition in build.sbt to use a newer rocksdbjni version if m1 was selected for the build.
• Add compatible libraries for libsentencepiece_jni and _sentencepiece_tokenizer with respective conditions for loading

TODO:

• Update docs with user guide on how to use m1 jar.

Motivation and Context

How Has This Been Tested?

Screenshots (if appropriate):

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• Code improvements with no or little impact
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

[maziyarpanahi]

@DevinTDHa Can we not call those native files aarch64? (I am not sure if these files are compatible with Linux aarch64 which Databricks/AWS are offering now, so when we start supporting those it maybe confusing - but if they are compatible with all aarch64 and not just M1 we can leave it like this, otherwise let's replace aarch64 with m1)

[DevinTDHa]

@DevinTDHa Can we not call those native files aarch64? (I am not sure if these files are compatible with Linux aarch64 which Databricks/AWS are offering now, so when we start supporting those it maybe confusing - but if they are compatible with all aarch64 and not just M1 we can leave it like this, otherwise let's replace aarch64 with m1)

Yes, let me change it actually to M1 so it is clearer. Regarding Linux and Windows compatibility there shouldn't be an issue, as this is for the mac dylibs only (i'll add another condition for SentencePieceJNI.java though)

[coveralls]

Coverage increased (+0.02%) to 74.376% when pulling 4b0dede on DevinTDHa:m1_support into 5ea6c91 on JohnSnowLabs:release/400-release-candidate.