[Doc] better documentation about credentials and service accounts.

[HailongWen]

This is a feature request to write better documentation on the credential connection property for Spanner JDBC Driver.

It is based on a question from Cloud Spanner customer:

With JDBC the string has the last parameter as a credentials file which appears to be a required parameter. On Kubernetes service accounts are used. Do we need a credentials file when using a Service Accounts with Kubernetes? Really the question is when/how to use a credentials file versus when/how to use a Service Account?

The answer is given by Shanika:

The credentials file is an optional parameter that can be used to provide specific credentials to use with the JDBC connection. The supported authentication methods with the JDBC driver are:

• Specify a credentials file in the connection URL (or connection properties): Both a user account credentials file as well as a service account credentials file can be used for this. Credentials files for service accounts can be downloaded from the IAM section of the project console in Google Cloud.
• Specify a valid OAuth token in the connection URL: If the application somehow already has authenticated against Google Cloud and has a valid OAuth token, that OAuth token can be included in the connection URL.
• Use the default authentication of the environment: If you do not specify any credentials file or OAuth token in the connection URL, the JDBC driver will use the default Google Cloud authentication of the environment. This is the same as using the normal Spanner client library without setting any credentials on the SpannerOptions. The credentials returned by the static method GoogleCredentials#getApplicationDefault() is what will be used in that case.

The existing documentation does not explicitly talk about these details. It would be nice to provide the information above in the Javadoc for JdbcDriver or in the README.