Trace context not propagated in dynamic WebSocket input

[NAIM6246]

Hello,
We’re currently implementing distributed tracing in our system and ran into an issue with dynamic WebSocket input. We have an upstream service sending messages to redpanda-connect over dynamic websocket input. These messages already include trace context (traceparent / tracestate).
The problem here is when messages arrives via the dynamic websocket input:

• Benthos does not continue the existing trace context.
• Instead, it starts a new root trace.
  As a result, we end up with two separate traces:

1. One from the upstream service → until Benthos
2. Another from Benthos → downstream services
   These traces are not linked, breaking end-to-end observability.

[anasya271-glitch]

hi everyone! i am a data engineer looking to contribute. i agree with this issue. distributed tracing context (traceparent and tracestate) should be propagated through the dynamic websocket input rather than starting a new root trace, as breaking end-to-end lineage hurts pipeline observability. could a maintainer please assign this issue to me? i would love to dive into the websocket input logic to see where the incoming headers are dropped and open a pull request to fix it.