First implementation for DRIVERS-2757.

Use Case

As a... Node.js driver user
I want... DNS lookup failures to be retried automatically on the driver side
So that... I don't have to build retry logic into my application

User Experience

• Less frequent connection failure errors

Dependencies

• None

Risks/Unknowns

• What could go wrong while implementing this change? (e.g., performance, inadvertent behavioral changes in adjacent functionality, existing tech debt, etc)
• Which actual errors should we retry? Only transient ones or all?
  • https://nodejs.org/api/dns.html#error-codes
  • Check to find out what the error codes mean
  • Only TIMEOUT, all other errors indicate something that is unrecoverable
• Is there an opportunity to improve existing documentation on this subject?
  • Users experience the DNS error at the call site for client.connect() so we'll update the API docs there. And there will be release notes for this.

Acceptance Criteria

Implementation Requirements

• Retry the initial srv txt lookups once on failure: TIMEOUT

Testing Requirements

• Use the dns seedlist spec prose test SRV records to hit a real DNS record and use sinon to fake timeout errors.
  • Note: try to write the testing in a way that is potentially adaptable as a general drivers prose test. If drivers have a DNS stubbing system they can do this, otherwise, this isn't something that can be made common easily.

Documentation Requirements

• Update client.connect() API doc.

Follow Up Requirements

• N/A, DRIVERS-2757 is the follow up