Elastic APM reporting seems to drop spans after a maximum per HTTP endpoint call

[iansltx]

💥  Actual behavior

When looking at a trace for an HTTP call, Elastic APM is missing later spans in the trace when a trace has a bunch of spans (e.g. for N+1 DB calls in a list endpoint).

🧑‍💻  Steps to reproduce

1. Turn Elastic APM tracing on
2. Enroll 100 hosts
3. Call GET /hosts?populate_software=true

🕯️ More info

As of creating this issue, I haven't confirmed whether OpenTelemetry has the same issue, so using OTel as a sink for traces rather than Elastic APM may be a near-term workaround for long traces.