fix(sdk-crash): Add Trace ID

[thetruecpaul]

We strip SDK crash errors in postprocess. This strips out their Trace ID; we need Trace IDs to store errors in EAP.

This PR adds a new Trace ID to these errors.

[sentry]

Bug: The code generates a trace_id but sets span_id to None, which is inconsistent and may cause issues in downstream systems that expect a valid span_id.
_{Severity: MEDIUM}

Suggested Fix

Generate a valid span_id in addition to the trace_id. Instead of setting span_id to None, create a 16-character hex string, for example: "span_id": uuid4().hex[:16]. This ensures the trace context is complete and consistent.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent. Verify if this is a real issue. If it is, propose a fix; if not, explain why it's
not valid.

Location: src/sentry/utils/sdk_crashes/sdk_crash_detection.py#L134-L136

Potential issue: The code adds a trace context to SDK crash events by generating a
`trace_id` but explicitly sets the corresponding `span_id` to `None`. While the event
schema allows a null `span_id`, this is inconsistent with other parts of the codebase
where a valid hex `span_id` is always generated alongside a `trace_id`. This could cause
issues for downstream consumers like the EAP system, which may expect both a valid
`trace_id` and `span_id` to be present for tracing. A trace context with a valid trace
ID but a null span ID is semantically unusual and may be rejected by tracing
infrastructure.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

[cleptric]

Maybe it makes sense to fix this in the event manager. I imagine there could be more places that create an event that could lack a trace ID.

[thetruecpaul]

We handle most of them in Relay — this appears to be unique among error-type occurrences in that it strips that relay-set default in postprocess. (Confirmed by looking at logs re: which projects are missing trace ID — only related to this case, it seems.)

Might be worth adding there as another backstop if this doesn't work, though.