[telemetry] Clean up cloud provider usage collector & add retry logic.

[lukeelmers]

In #95131, we migrated some legacy "cloud detector" logic from the monitoring plugin into kibana_usage_collection in order to collect some basic information about what cloud environment Kibana is running in via a new cloud_provider usage collector. (src/plugins/kibana_usage_collection/server/collectors/cloud)

In reviewing the old code, we identified a few enhancements & cleanup tasks we'd like to address, some of which we didn't do at the time as it would have required some more heavy refactoring:

1. getting rid of the constructor args which are only there for testing purposes (request, fs), and mocking those in the tests
2. getting rid of the use of the request lib in favor of something like node-fetch which we use more widely in kibana (request looks like it's mostly used in tests atm, not many other places)
3. adding some retry logic to the detector in the event of a retryable failure[0]
4. updating GCP logic to still return values even if a request for one of the fields fails, instead of using an all-or-nothing approach

[0] Some of the API docs have info on possible error codes that can be returned, which we should take a look at when determining our retry logic:
https://docs.microsoft.com/en-us/azure/virtual-machines/windows/instance-metadata-service
https://cloud.google.com/compute/docs/storing-retrieving-metadata
https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instancedata-data-retrieval.html

[elasticmachine]

Pinging @elastic/kibana-core (Team:Core)