fix: struct metadata parsing for nested decimal fields

[sshpuntoff]

Description

When parsing Struct<Decimal(10,20)>, the driver currently raises an exception:

  java.lang.ArrayIndexOutOfBoundsException: Index 1 out of bounds for length 1
      at com.databricks.jdbc.api.impl.MetadataParser.parseStructMetadata(MetadataParser.java:30)
      at com.databricks.jdbc.api.impl.ComplexDataTypeParser.parseToStruct(ComplexDataTypeParser.java:104)
      at com.databricks.jdbc.api.impl.ComplexDataTypeParser.convertValueNode(ComplexDataTypeParser.java:131)
      at com.databricks.jdbc.api.impl.ComplexDataTypeParser.parseToArray(ComplexDataTypeParser.java:73)

This occurs because the current implementation does not expect nested comma's in datatypes.
By tracking the Depth of the () characters, we can successfully parse complex datatypes.

Issue:
#981

Testing

Unit tests expanded and local testing done.

Additional Notes to the Reviewer

This is a bug in the implementation of the MetaDataParser and is pretty direct to re-produce and correct.

[sshpuntoff]

there is a case to be made that just a single
int depth = 0; would work correctly here, and the code below just needs to be

if (ch == '<' || ch == '(') {
 depth ++;
}
else if (ch == '>' || ch == ')') {
    depth --;
}

Which would simplify the change set even more.
Let me know if you prefer a minimal diff over the explicit variable names.