refactor: use typed tries in trace_decoder

[0xaatif]

#275

Change summary:

• Add typed_mpt::{StateTrie, StorageTrie, ..}.
• type1 frontend emits the above.
• type1 frontend is cleaner.

[0xaatif]

Drafting to wait for #400

[muursh]

I'm good with this. I have one question though. Have you tried benchmarking this vs the old version?

[0xaatif]

I have one question though. Have you tried benchmarking this vs the old version?

No, but

zk_evm/trace_decoder/src/lib.rs

Lines 16 to 17 in a5b92b2

	//! # Non-Goals
	//! - Performance - this won't be the bottleneck in any proving system.

If I had to guess, worst case we've got constant overhead, but I think this mostly just changes where rlp::decode/rlp::encode calls (to behind the type)

[muursh]

I have one question though. Have you tried benchmarking this vs the old version?

No, but

zk_evm/trace_decoder/src/lib.rs

Lines 16 to 17 in a5b92b2

	//! # Non-Goals
	//! - Performance - this won't be the bottleneck in any proving system.

If I had to guess, worst case we've got constant overhead, but I think this mostly just changes where rlp::decode/rlp::encode calls (to behind the type)

Yeah it's not a requirement to be more performant or anything, hence why I approved. I was just being nosey