Fix connection_retries to override retry_limit

[samshinde]

Describe the bug

While trying to override retry_limit with connection_retries to halt the connection attempts with limited no of time and blow up in case of any issue, due to improper merging, the connection_retires (limited number of attempts) never sets and the program continues to run until default retry_limit(mostly 200) causing user to hold for a longer period of time.

To Reproduce
Steps to reproduce the behavior:

Pass an incorrect password in the bootstrap command.

knife bootstrap 40.x.x.x --connection-user azure --connection-password password --connection-protocol winrm  -N test-node

Expected behavior

It should raise an error after a few failed attempts to connect.

Version Information:

• chef: 15.2.29
• inspec: 4.10.4
• winrm: 2.3.2
• train-winrm: 0.2.4

Additional context
As per the analysis, this can easily be fixed by correcting the merge order in opts in the following way.

        opts = {
          retry_limit: @connection_retries.to_i,
          retry_delay: @connection_retry_sleep.to_i,
        }
        opts = retry_options.merge(opts)

[kekaichinose]

@btm can I get your eyes on this?

[kvivek1115]

@btm & @clintoncwolfe could you please have a look the PR #19 ?