Support AWS credential profiles in bottomless S3 replication ##

Currently, libsql's bottomless S3 replication feature only accepts AWS credentials through environment variables:

• LIBSQL_BOTTOMLESS_AWS_ACCESS_KEY_ID
• LIBSQL_BOTTOMLESS_AWS_SECRET_ACCESS_KEY
• LIBSQL_BOTTOMLESS_AWS_SESSION_TOKEN (optional)
• LIBSQL_BOTTOMLESS_AWS_DEFAULT_REGION

The implementation does not read from the standard AWS credentials file (~/.aws/credentials) and doesn't support specifying a particular AWS profile.

Problem

This limitation creates several issues:

1. It doesn't align with AWS best practices of using credential files and profiles
2. It requires users to explicitly set environment variables even when credentials are already configured
3. It makes it difficult to use in environments with multiple AWS accounts/roles
4. It doesn't leverage the AWS SDK's built-in credential chain capabilities

Proposed Solution

Enhance the AWS credential handling in bottomless S3 replication to:

1. Follow the standard AWS SDK credential provider chain
2. Add support for reading from ~/.aws/credentials
3. Add an optional environment variable to specify which profile to use
4. Fall back to environment variables if no profile is specified or found

Start by locating the bottomless S3 replication credential-handling entry point and reviewing how the listed LIBSQL_BOTTOMLESS_AWS_* variables are consumed. Compare that behavior with the AWS SDK credential provider chain and determine the intended profile environment variable. Done means credentials can be loaded from the standard AWS credentials file, a selected profile is supported, and the existing environment-variable behavior remains available.