feat: Support proxy for debugger endpoint v2

[lym953]

Now we support proxying requests from /debugger/v1/input. This PR adds /debugger/v2/input and /debugger/v1/diagnostics.

#925

Test

Steps

1. Set up Exception Replay following https://docs.datadoghq.com/error_tracking/backend/exception_replay/
2. Build a test layer and install it on the Lambda
3. Set DD_TRACE_DEBUG to true
4. Change Lambda timeout from 3s to 30s. Seems Exception Replay dramatically increases the duration of the first invocation. It took 8–9s for my tests.

Result

Before:

• In CloudWatch logs, see the error multiple times: debugger::unsupported_agentUnsupported Datadog agent detected. Snapshots from Dynamic Instrumentation/Exception Replay/Code Origin for Spans will not be uploaded. Please upgrade to version 7.49.0 or later

After:

• No such error.
• See tracer debug log: Detected /debugger/v2/input endpoint
• See the error on Error Tracking page

Notes

Thanks @nhulston @joeyzhao2018 @purple4reina for discussion and helping debug.

[lym953]

Is it correct to use Self::debugger_logs_proxy? Should any param here be different?

datadog-lambda-extension/bottlecap/src/traces/trace_agent.rs

Lines 388 to 398 in 0a5e11e

	async fn debugger_logs_proxy(State(state): State<ProxyState>, request: Request) -> Response {
	Self::handle_proxy(
	state.config,
	state.proxy_aggregator,
	request,
	"http-intake.logs",
	DEBUGGER_LOGS_INTAKE_PATH,
	"debugger_logs",
	)
	.await
	}

@nhulston

[nhulston]

Sorry, I was incorrect in Slack. After looking at the datadog-agent code, it looks like the params changed in v2. See the suggested code changes in the other comments

[nhulston]

Suggested change

	const V1_DEBUGGER_LOGS_INTAKE_PATH: &str = "/api/v2/logs";
	const V2_DEBUGGER_INTAKE_PATH: &str = "/api/v2/debugger";

[nhulston]

Looks good. Can we test in a real Lambda before merging?

[lym953]

Tested. Will merge after v89 is released.