After Confluence upgrade, The duplicate index and incorrect data type caused MSSQL Database CPU spike

Issue Summary

Upgrading to 6.2.3 causes a duplicate index on the BANDANA table named `band_cont_key_idx` with column "BANDANACONTEXT", "BANDANAKEY". This currently affects SQL Server database.

This upgrade also causes the BANDANA table column "BANDANACONTEXT", "BANDANAKEY" to have the `varchar` type instead of the intended type `nvarchar`.

Steps to Reproduce

Upgarding Confluence to the version 6.2.3 with MSSQL database.

Expected Results

The query against the bandana table should neither long-running query or high CPU contributor.

Actual Results

A simple select from bandana table is contributing around 20% database server CPU or become long-running query.

Workaround

Manually change the column type to nvarchar with the following query:

ALTER TABLE BANDANA DROP CONSTRAINT bandana_unique_key;
DROP INDEX BANDANA.band_cont_key_idx;
DROP INDEX BANDANA.band_context_idx;
ALTER TABLE BANDANA ALTER COLUMN BANDANACONTEXT NVARCHAR(255) NOT NULL;
ALTER TABLE BANDANA ALTER COLUMN BANDANAKEY NVARCHAR(100) NOT NULL;
CREATE NONCLUSTERED INDEX band_cont_key_idx ON BANDANA(BANDANACONTEXT, BANDANAKEY);
CREATE NONCLUSTERED INDEX band_context_idx ON BANDANA(BANDANACONTEXT);
ALTER TABLE BANDANA ADD CONSTRAINT bandana_unique_key UNIQUE (BANDANACONTEXT, BANDANAKEY);