* Remove unused submodule output

This breaks plan when trying to create a read replica without setting
the master password.

module.database.module.db_instance.local.this_db_instance_password: local.this_db_instance_password: Resource 'aws_db_instance.this' does not have attribute 'password' for variable 'aws_db_instance.this.*.password'

* Add postgres and mysql replica example

* Add tags to enhanced_monitoring iam_role

* give priority to tags defined by user

* Create parameter group with custom name

* Allow setting of parameter group description too

* Move locals into main.tf

* Include "no_prefix" resource in output

* Update all README files with terraform docs

* Add-deletion_protection-argument

* Add-deletion_protection-argument

Add ability to set terraform timeouts{} map for aws_db_instance resources

* Support database option group

* support option group with latest

* adding two missing arguments for aws_db_instance: timezone & character_set_name

* added character_set_name to oracle example

* added example for mssql

* added link to readme

* moved tags to last line

* fixed mssql example

* Added conditional mssql db_instance based on var.engine. Added conditional outputs

* ✨ Enable db_subnet_group_name variable

If the variable not specify then use the db_subnet_group module

* Disable create_db_subnet_group if db_subnet_group_name is defined

* Support conditional creation for the database too.

* Database instance outputs need to accommodate the case where count is zero.

* Declare variables that allow conditional creation of the database, parameter group, and subnet group. Pass them from main.tf to the nested modules.

* Consistently accommodate boolean like arguments.

aws_db_parameter_group (or AWS?) does not support removing previously
defined parameters.

I thought I'd give "taint" a spin and force the parameter group to be
recreated.  terraform tried to delete the parameter group but failed as
it's still in use.

This change allows terraform to migrate the RDS instance to the new
parameter group before destroying the old.

Accessing aws_db_subnet_group.this when its count = 0 is an error, so we
need to instead access it via splat syntax and then apply a default value
for when the resulting list is empty.