Performance killer: SQL query using nvarchar instead of char

[shubmitt]

This issue has been posted here. I believe it has carried over from this library. So sharing here also.

ESSolutions#50

An example query where conversion is happening -

(@p3 nvarchar(64),@P4 nvarchar(64),@p1 int,@p2 datetime2(7))UPDATE [transactions_transaction] SET [is_split_transaction] = @p1, [modified] = @p2 WHERE ([transactions_transaction].[transaction_id] = @p3 AND [transactions_transaction].[user_id] = @P4)

both user_id and transaction_id are CustomUUIDField (char(32)) and we dont want pyodbc to convert this params