Add retries when connecting to backend database

[dgajendr]

Is your feature request related to a problem? Please describe.
Currently, our great expectations is set up with AWS RDS on backend. Occasionally, the connection gets lost failing the entire job.

Describe the solution you'd like
Please add retry to the main great expectations class, so the number of retries to connect to backend on failure can be specified.

Describe alternatives you've considered
Retrying the whole job, but it resulted in other side effects