Add Environment & Services support

[MikeGoldsmith]

Which problem is this PR solving?

Adds E&S support by ensuring service name and dataset are configured correctly for both classic and non-classic API keys. Classic keys use the dataset and non-classic uses the servicename for target dataset.

The beeline also now assigns a default service name based on the API key provided. Classic gets "nodejs" and non-classic gets "unknown_service:{process-name}" where process name is the node process.name name if assigned and falls back to "nodejs" is it cannot resolve one.

The Beeline now applies the following defaults:

Property	Classic	Non-classic
service name	unknown_service:{process.name|nodejs}	unknown_service:{process.name|nodejs}
dataset	nodejs	unknown_service

Note, I selected to not update the default dataset from nodejs to beeline-nodejs as this would be a break change for anyone relying on that now. We can make the change in the next major release.

• Closes Update for E&S #550

Short description of the changes

• Update libhoney.configure to detect and apply defaults for service name and dataset
• Warn on missing API key, dataset (classic) and service name (non-classic)
• Update tests and add extra tests for default service name and dataset
• Don't propagate dataset in Honeycomb propagation implementation (w3c doesn't propagate dataset field)

[JamieDanielson]

Confirmed this works with EGS 👍 ... but it looks like we still need to add service.name.

Maybe add something like exports.TRACE_SERVICENAME = "service.name"; in schema.js (since the underscore variable is already used), and [schema.TRACE_SERVICENAME]: opts.serviceName, in libhoney.js? Then quick add to index.test.js for libhoney default config tests.

[JamieDanielson]

Nice! As we chatted about, can consider using Template literals for the string concatenation, but otherwise this looks great.

[JamieDanielson]

🚀