Improve `debug_traceBlock` performance

[mattsse]

Describe the feature

currently, trace_block is a loop over transactions and that invoke trace_transaction:

https://github.com/paradigmxyz/reth/blob/08bdec9efa26c6a97a2c0467469727b61c9b5dfc/crates/rpc/rpc/src/debug.rs#L107-L118

note how this clones the opts:

https://github.com/paradigmxyz/reth/blob/08bdec9efa26c6a97a2c0467469727b61c9b5dfc/crates/rpc/rpc/src/debug.rs#L119-L119

this results in some overhead:

• opts -> TracingConfig conversion https://github.com/paradigmxyz/reth/blob/08bdec9efa26c6a97a2c0467469727b61c9b5dfc/crates/rpc/rpc/src/debug.rs#L708-L710
• Inspector is not reused: for block level tracing we can reuse the Inspector by fusing it https://docs.rs/revm-inspectors/latest/revm_inspectors/tracing/struct.TracingInspector.html#method.fuse and because most geth trace results don't consume the traces, we can keep the allocated memory in TracingInspector

https://docs.rs/revm-inspectors/latest/revm_inspectors/tracing/struct.TracingInspector.html#method.geth_builder

TODO

• this should be improved in such a way that the allows us to reuse the internals

this could perhaps be done with a helper type for block tracing that keeps the TracingInspector or similar

Additional context

No response

[JosepBove]

I can take this

[JosepBove]

Ready for review