Implement retry logic in read requests for services and service integrations

[mLupine]

Ever since we started using Terraform to manage our PagerDuty configuration at @nordcloud, we've been facing an issue with state refreshing calls failing during the execution of the terraform plan command.

Our configuration contains a few thousand services and service integrations, and almost every time the command was ran, it (probably) started reaching the PD API webserver's rate limits, failing with the following error messages:

Error: Error reading: PXXXXXX: Get https://api.pagerduty.com/services/PXXXXXX: http2: server sent GOAWAY and closed the connection; LastStreamID=1999, ErrCode=NO_ERROR, debug=""

Error: Error reading: PXXXXXX: Get https://api.pagerduty.com/services/PXXXXXX: http2: server sent GOAWAY and closed the connection; LastStreamID=1999, ErrCode=NO_ERROR, debug=""

After some debugging, I found out that if the failing calls are executed again, even with no delay, they actually go through without any issues.

This change introduces Terraform's retry mechanism to the "Read" command of pagerduty_service and pagerduty_service_integrations, making it return RetryableError every time an HTTP request to the API server fails with a status code other than 404.

I've implemented the updated provider in our CI/CD flow and it's been running flawlessly ever since.

Please keep in mind that it's my first ever encounter with Golang, so don't be too harsh on me if the code isn't perfect 😉

[stmcallister]

Looks great! Nice job! Thanks for contributing this!