Fix creating unique index on non-unique field

[fkollmann]

What did this pull request do?

When having a named unique index on an non-unique column, the migrator will not check, if the index already exists, resulting in Error 1061 (42000): Duplicate key name 'index_name'

This fix adds the check, as done in the unique column code branch of the function (line 101).

User Case Description

Creating an index on a non-unique column, but not having a unique constraint:

type Ticket struct {
  ID        uint64            `gorm:"primarykey"`
  Hashes    datatypes.JSON    `gorm:"column:hashes;uniqueIndex:uq_tickets_hashes,expression:(CAST(hashes AS CHAR(40) ARRAY))"`
}

On the first migration, the index is successfully being created.

On the next migration the following error is given, and no information_schema check is being performed:

error="Error 1061 (42000): Duplicate key name 'uq_tickets_hashes'" sql="CREATE UNIQUE INDEX `uq_tickets_hashes` ON `tickets__simple`((CAST(hashes AS CHAR(40) ARRAY)))"
    parser.go:150: failed to migrate database: failed to migrate tickets table: Error 1061 (42000): Duplicate key name 'uq_tickets_hashes'

After the fix, the check is being performed:

sql="SELECT count(*) FROM information_schema.statistics WHERE table_schema = 'unittest' AND table_name = 'tickets__simple' AND index_name = 'uq_tickets_hashes'"