Allow backend DB to authenticate using temporary tokens

### Description Based on [this](https://github.com/apache/airflow/discussions/30283) discussion. Currrently there is no way to use token identity to authenticate with amazon RDS without a fairly significant change to the helm charts and airflow code. I will implement this functionality and add the helm options as: ```yaml externalDatabase: type: postgres host: airflow-cluster.<uniqueId>.us-east-1.rds.amazonaws.com ## the port of the external database ## port: 5432 ## the database/scheme to use within the external database ## database: airflow ## the username for the external database ## user: airflow awsRdsTokenIdentity: enabled: true region: us-east-1 connectionExpirySeconds: 600 ``` And use sqlalchemy envents to provide the token. ```python def amend_connection(cparams): if conf.getboolean("database", "use_aws_token_identity"): log.info(f'connecting user {cparams["user"]} to {cparams["host"]}:{cparams["host"]} using pod identity') client = boto3.client( "rds", region_name=conf.get_mandatory_value("database", "aws_region"), ) token = client.generate_db_auth_token( DBHostname=cparams["host"], Port=cparams["port"], DBUsername=cparams["user"], ) cparams["password"] = token else: log.info(f'connecting {cparams["user"]} using user/password') @event.listens_for(engine, "do_connect") def provide_token(dialect, conn_rec, cargs, cparams): amend_connection(cparams) ``` ### Use case/motivation Temporary credentials are a security feature generally required secops and a general good practice these days, so it makes sense for me to support them. ### Related issues _No response_ ### Are you willing to submit a PR? - [X] Yes I am willing to submit a PR! ### Code of Conduct - [X] I agree to follow this project's [Code of Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)