Protocol now required when using `custom.localstack.host`

Description

Since serverless-loaclstack v1.2.1 it appears that the protocol needs to be included in the custom.localstack.host property.
In older versions it appears this was not required.

Configuration

In our serverless configuration we had something similar to:

custom:
  localstack:
    host: ${env:SOME_VAR, '127.0.0.1'}

i.e. without the protocol (http://).

serverless-localstack v1.2.0:

Using this with:

• serverless: v3.36.0 (using an older version to be able test with older plugin versions)
• serverless-localstack: v1.2.0

The debug output shows:

Reconfiguring service acm to use http://127.0.0.1:4566
Reconfiguring service amplify to use http://127.0.0.1:4566
Reconfiguring service apigateway to use http://127.0.0.1:4566

serverless-localstack v1.2.1:

Using this with:

• serverless: v3.36.0 (using an older version to be able test with older plugin versions)
• serverless-localstack: v1.2.1

The debug output shows:

Reconfiguring service acm to use 127.0.0.1://127.0.0.1:4566
Reconfiguring service amplify to use 127.0.0.1://127.0.0.1:4566
Reconfiguring service apigateway to use 127.0.0.1://127.0.0.1:4566

which then results in errors such as:

aws: [1] { UnknownEndpoint: Inaccessible host: `127.0.0.1' at port `443'. This service may not be available .....

Additional information

In the documentation the example does include the protocol when setting the custom.localstack.host so I suppose it can also be seen as a configuration error on our end.. but in that case a clearer error message might be useful.

I'm also guessing the change in behavior is caused by #243

Reproduce the reported configuration with serverless-localstack v1.2.0 and v1.2.1, then inspect the change referenced in #243 and the code that builds the LocalStack endpoint. Done should mean a host without a protocol either continues to work as before or produces a clear configuration error, with regression coverage for the behavior.