Support pprof format natively

The pyroscope ecosystem is around Go pprof. I think it makes sense that the agent would be able to emit pprof natively in order to improve data flows that would operate on the canonical form. It can make them simpler and more efficient. For example:

- it will make writing tools in different languages easier with the oss pprof support, as I only see async-profiler java jfr2pprof (which you use in [JPProf](https://github.com/grafana/JPProf)) and your Go server parser for JFR. I have a node app (unfortunately) where I need to simulate some pyroscope api and I would like to receive pprof
- apply a more efficient format early
- provide flexibility to tradeoff parse performance overhead across the pipeline depending on the use case, for example offload large scale parse work to the distributed data plane where it may be more negligible.

I think this should be complementary to an agent capability https://github.com/grafana/alloy/issues/395

Inizia esaminando come il Java agent emette attualmente i dati di profiling e come è strutturata la sua integrazione con la Pyroscope API; l'issue non indica file né test. Confronta l'output nativo Go pprof desiderato con l'issue di Alloy collegata e i riferimenti JPProf/JFR esistenti, quindi definisci il percorso di output supportato e la validazione per dati pprof canonici.

Scritto dal modello di indicizzazione a partire dal testo della issue.